feat(proxy): add TypeSafe AI Jev evaluate passthrough with registry-priced spend tracking by devin-ai-integration[bot] · Pull Request #41607 · BerriAI/litellm

devin-ai-integration · 2026-09-17T15:57:17Z

TLDR

Problem this solves:

Jev (TypeSafe AI) has no path through the gateway
Its evaluate API is not chat completions, so no provider fits
Teams cannot use a virtual key or track Jev spend

How it solves it:

New /typesafe/{endpoint} passthrough to api.typesafe.ai
Proxy injects TYPESAFE_API_KEY, callers use virtual keys
Spend priced from the model registry using the response usage
Registry entries for jev-1.13.0, jev-latest, jev-preview
Docs: https://docs.litellm.ai/docs/pass_through/typesafe (litellm-docs Langsmith: Add envs for project/run names; fix bug with None metadata #1524, merged)

User Flow

Before: a developer who wants Jev's structured decisions behind their gateway key has nowhere to send the request

The proxy admin sets TYPESAFE_API_KEY and restarts the proxy
The developer sends POST https://litellm-domain/typesafe/v1/systemone with their virtual key and {"state": "...", "model": "jev-latest", "questions": {...}}
They get a 404 {"detail":"Not Found"}, the route does not exist
They fall back to calling api.typesafe.ai directly with the raw TypeSafe key, outside key auth, budgets, and spend logs

After: the same request goes through the gateway and shows up in spend logs

The proxy admin sets TYPESAFE_API_KEY (and optionally TYPESAFE_API_BASE) and restarts the proxy
The developer sends POST https://litellm-domain/typesafe/v1/systemone with their virtual key and the same body
They get TypeSafe's 200 response unchanged: model, answers with choice, probabilities, confidence, and usage
https://litellm-domain/ui/?page=logs shows the request as typesafe/jev-1.13.0 with the input token count and spend at the registry input price

Relevant issues

Affected release

Linear ticket

Resolves LIT-7975

Pre-Submission checklist

Please complete all items before asking a LiteLLM maintainer to review your PR

I have added meaningful tests
The handful of test files covering my change pass locally, e.g. uv run pytest tests/test_litellm/<your_test_file>.py -v. Leave the suites (make test-unit-*, make test-unit) to CI: it finishes in ~15 minutes where a laptop takes an hour or more
My PR passes all required CI/CD checks (e.g., lint, schema.d.ts sync check, etc.)
My PR's scope is as isolated as possible; it only solves 1 specific problem
I have received a Greptile Confidence Score of at least 4/5 before requesting a maintainer review (Greptile reviews automatically once the PR is opened; only comment @greptileai to re-request a review after pushing changes)

Delays in PR merge?

If you're seeing a delay in your PR being merged, ping the LiteLLM Team on Slack (#pr-review).

Screenshots / Proof of Fix

Topology on both legs: proxy_cli.py --num_workers 2 --use_v2_migration_resolver booted from the named commit, one local Postgres database per leg, the live api.typesafe.ai upstream (real spend), and a virtual key from POST /key/generate with max_budget: 1. Same curls in the same order on both sides

Pricing evidence, parsed from https://docs.typesafe.ai/models.md on 2026-09-17: Jev 1.13 | jev-1.13.0 | Price (per Btok / per Mtok) | $42 / $0.042, "Charged per input token. Output tokens are free", aliases jev-latest -> jev-1.13.0 and jev-preview -> jev-1.13.0

Before (`87650bf`)

$ curl -sS -i -X POST http://127.0.0.1:27514/typesafe/v1/systemone -H "Authorization: Bearer $VIRTUAL_KEY" -H "Content-Type: application/json" -d '{"state":"Help! My payouts have been failing for 3 days.","model":"jev-latest","questions":{"department":{"type":"choice","instructions":"Which team should handle this?","criteria":{"billing":"Payments and invoices","technical":"Bugs, outages, integration errors","sales":"Pricing and plans"}}}}'
HTTP/1.1 404 Not Found
{"detail":"Not Found"}

$ curl -sS -i http://127.0.0.1:27514/typesafe/v1/models -H "Authorization: Bearer $VIRTUAL_KEY"
HTTP/1.1 404 Not Found
{"detail":"Not Found"}

$ curl -sS -i -X POST http://127.0.0.1:27514/typesafe/v1/systemone -H "Content-Type: application/json" -d '<same body>'
HTTP/1.1 404 Not Found
{"detail":"Not Found"}

The logs page has nothing to show for these calls:

After (`1feaa48`)

$ curl -sS -i -X POST http://127.0.0.1:21879/typesafe/v1/systemone -H "Authorization: Bearer $VIRTUAL_KEY" -H "Content-Type: application/json" -d '{"state":"Help! My payouts have been failing for 3 days.","model":"jev-latest","questions":{"department":{"type":"choice","instructions":"Which team should handle this?","criteria":{"billing":"Payments and invoices","technical":"Bugs, outages, integration errors","sales":"Pricing and plans"}}}}'
HTTP/1.1 200 OK
x-litellm-call-id: b88df75a-4be9-4138-adc5-2b651d7320bf
x-litellm-model-api-base: https://api.typesafe.ai/v1/systemone
{"model":"jev-1.13.0","answers":{"department":{"type":"choice","choice":"billing","confidence":0.73,"probabilities":{"technical":0.18,"billing":0.82,"sales":0.0}}},"usage":{"input_tokens":342,"output_tokens":38}}

$ curl -sS -i http://127.0.0.1:21879/typesafe/v1/models -H "Authorization: Bearer $VIRTUAL_KEY"
HTTP/1.1 200 OK
x-litellm-call-id: 8ad333ff-bb74-4871-b6f3-c669dc3292b8
{"models":[{"name":"jev-latest","description":"The latest iteration of TypeSafe's System One Model: Jev","release_date":"2026-09-10T18:38:01.391457+00:00"},{"name":"jev-preview","description":"A preview version of `jev-latest`: should be better in most ways","release_date":"2026-09-10T18:39:06.057655+00:00"}]}

$ curl -sS -i -X POST http://127.0.0.1:21879/typesafe/v1/systemone -H "Content-Type: application/json" -d '<same body>'
HTTP/1.1 401 Unauthorized
{"error":{"message":"Authentication Error, No api key passed in.","type":"auth_error","param":"None","code":"401"}}

Spend rows, read back through the spend API:

$ curl -sS "http://127.0.0.1:21879/spend/logs?request_id=b88df75a-4be9-4138-adc5-2b651d7320bf" -H "Authorization: Bearer $LITELLM_MASTER_KEY" | jq '[.[] | {model, custom_llm_provider, prompt_tokens, completion_tokens, total_tokens, spend, api_key_alias: .metadata.user_api_key_alias}]'
[
  {
    "model": "typesafe/jev-1.13.0",
    "custom_llm_provider": "typesafe",
    "prompt_tokens": 342,
    "completion_tokens": 38,
    "total_tokens": 380,
    "spend": 0.000014364,
    "api_key_alias": "jev-qa-tip"
  }
]

$ curl -sS "http://127.0.0.1:21879/spend/logs?request_id=8ad333ff-bb74-4871-b6f3-c669dc3292b8" -H "Authorization: Bearer $LITELLM_MASTER_KEY" | jq '[.[] | {model, custom_llm_provider, prompt_tokens, completion_tokens, spend}]'
[
  {
    "model": "typesafe/unknown",
    "custom_llm_provider": "typesafe",
    "prompt_tokens": 0,
    "completion_tokens": 0,
    "spend": 0.0
  }
]

342 * 4.2e-8 + 38 * 0 = 0.000014364, logged under the versioned model from the response rather than the jev-latest alias from the request

Dependents of the registry rows and the new route, driven on both sides with the same script:

Path	`87650bf`	`1feaa48`
`POST /model/new` with `typesafe/jev-latest` in `model_list`	500	500
`GET /v1/models` after that	not listed	not listed
`POST /v1/chat/completions` on that model	400 invalid model	400 invalid model
`GET /health?model=<it>`	0 healthy, 0 unhealthy	0 healthy, 0 unhealthy
`GET /model_group/info?model_group=<it>`	`[]`	`[]`
`GET /public/model_hub`	`[]` 200	`[]` 200
key with `allowed_routes: ["/typesafe/*"]`, `GET /typesafe/v1/models`	404	200
same key, `GET /v1/models`	403	403
`POST /model/delete`	200	200

Observations from the run:

Models listing logs typesafe/unknown, zero tokens: PR causes
POST /model/new with typesafe/jev-latest answers 500: PR leaves alone
x-litellm-model-api-base names the upstream URL: PR causes

Not verified: TYPESAFE_API_BASE pointing anywhere but the default host, a passthrough deployment credential instead of the env key, and a TypeSafe 4xx or 5xx flowing back through the route (the live API answered 200 to everything sent)

Type

🆕 New Feature

Caveats (if any)

Low

A call whose response names no model (GET /typesafe/v1/models) is logged as typesafe/unknown with zero tokens and zero spend, the same label the Anthropic and Gemini passthrough handlers use for that case. Alternatives considered: skip the spend row for such calls (loses the record of who used the route) or label it by request path (a label shape no dashboard reads). Fixing it is net-negative: it would add a third convention for one free call
The typesafe/* registry rows carry pricing only. POST /model/new with typesafe/jev-latest as a model_list entry answers 500 on both base and tip, so Jev is reachable through /typesafe/{endpoint} and nothing else. Pre-existing for any provider without a chat implementation, left alone here
mode: "evaluation" is a new registry mode added to the ModelInfoBase literal and the JSON schema. /model_group/info, /public/model_hub, and /v1/models never see it because no router entry can load these rows
Only TYPESAFE_API_KEY or a passthrough deployment credential reaches TypeSafe; the caller's own header is replaced, never forwarded, which is what makes virtual keys the only credential a client needs
The non-required osv-scan check fails on a soupsieve advisory (GHSA-gjv8-xp57-g29c, GHSA-j934-xhv5-fg8f) in a dependency this PR does not touch; build(deps): bump soupsieve from 2.8.4 to 2.9.2 #41679 bumps it

Final Attestation

The tests check the right things, including the edge cases, and regressions in the respective real-world customer use-cases are not possible after this PR

Link to Devin session: https://app.devin.ai/sessions/b8ce9e101c644969abb9bcc9dfa2c333
Open in Devin Desktop: https://app.devin.ai/desktop/session/b8ce9e101c644969abb9bcc9dfa2c333?variant=devin
Requested by: @mateo-berri

Note

Low Risk
Additive passthrough and logging path; upstream auth uses configured credentials only, with no changes to core chat routing or existing providers.

Overview
Adds TypeSafe AI (Jev) as a first-class proxy passthrough so clients can call evaluate endpoints with virtual keys instead of hitting api.typesafe.ai directly.

The proxy exposes GET/POST /typesafe/{endpoint:path}, forwards to TYPESAFE_API_BASE (default https://api.typesafe.ai) with TYPESAFE_API_KEY on the upstream Authorization header, and registers the path in gateway allowlists, lazy passthrough loading, and OpenAPI/UI schema.

Spend logging parses TypeSafe responses for model and usage.input_tokens / output_tokens, prices input from the model registry (output cost is zero for the new typesafe/jev-* entries), and logs calls as typesafe/{model} with fallback to the request model or unknown. Registry and schema gain a new evaluation model mode for these entries.

^{Reviewed by Cursor Bugbot for commit 1feaa48. Bugbot is set up for automated code reviews on this repo. Configure here.}

1feaa48 passes /live-pr-risk

Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>

devin-ai-integration · 2026-09-17T15:57:21Z

🤖 Devin AI Engineer

I'll be helping with this pull request! Here's what you should know:

✅ I will automatically:

Address comments on this PR. Add '(aside)' to your comment to have me ignore it.
Look at CI failures and help fix them

Note: I can only respond to comments from users who have write access to this repository.

⚙️ Control Options:

Disable automatic comment, CI, and merge conflict monitoring

codspeed · 2026-09-17T15:59:24Z

Merging this PR will not alter performance

✅ 31 untouched benchmarks

_{Comparing litellm_typesafe_passthrough (1feaa48) with main (356b8d4)}

Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>

greptile-apps · 2026-09-17T16:02:46Z

Confidence Score: 5/5

The PR appears safe to merge with no outstanding correctness, security, or repository-rule violations.

Findings

Schema Rejects Evaluation Mode ▶
POST Query Parameters Dropped ▶
Provider Code Outside llms ▶

Summary

Adds a TypeSafe AI Jev passthrough backed by virtual-key authentication and registry-priced spend tracking.

Registers authenticated GET and POST routes under /typesafe/{endpoint}.
Injects configured TypeSafe credentials while preserving the shared passthrough request path.
Normalizes Jev usage and model metadata for standard spend logging.
Adds input-only registry pricing for versioned, latest, and preview Jev models.
Updates generated API schemas and adds route, pricing, logging, and metadata tests.

_{Reviews (7) · Last reviewed commit: "fix(proxy): log TypeSafe calls that name..."}

greptile-apps · 2026-09-17T16:02:49Z

+    "typesafe/jev-1.13.0": {
+        "input_cost_per_token": 4.2e-08,
+        "litellm_provider": "typesafe",
+        "mode": "evaluation",


Schema Rejects Evaluation Mode

The new evaluation value is absent from the committed schema, so the required cost-map guard fails until it is regenerated

greptile-apps · 2026-09-17T16:02:49Z

+@router.api_route(
+    "/typesafe/{endpoint:path}",
+    methods=["GET", "POST"],
+    tags=["TypeSafe AI Pass-through", "pass-through"],
+)
+async def typesafe_proxy_route(
+    endpoint: str,
+    request: Request,
+    fastapi_response: Response,
+    user_api_key_dict: UserAPIKeyAuth = Depends(user_api_key_auth),
+):
+    """[Docs](https://docs.litellm.ai/docs/pass_through/typesafe)"""
+    base_target_url: Final = get_secret_str("TYPESAFE_API_BASE") or "https://api.typesafe.ai"
+    encoded_endpoint: Final = httpx.URL(endpoint).path
+    normalized_endpoint: Final = encoded_endpoint if encoded_endpoint.startswith("/") else f"/{encoded_endpoint}"
+    base_url: Final = httpx.URL(base_target_url)
+    updated_url: Final = base_url.copy_with(
+        path=HttpPassThroughEndpointHelpers.join_base_and_endpoint_path(base_url, normalized_endpoint),
+        params=request.query_params,
+    )
+    typesafe_api_key: Final = passthrough_endpoint_router.get_credentials(
+        custom_llm_provider="typesafe",
+        region_name=None,
+    )
+    endpoint_func: Final = create_pass_through_route(
+        endpoint=endpoint,
+        target=str(updated_url),
+        custom_headers={
+            "Authorization": f"Bearer {typesafe_api_key}",
+            "Content-Type": "application/json",
+        },
+        custom_llm_provider="typesafe",
+        is_streaming_request=False,
+    )
+    return await endpoint_func(request, fastapi_response, user_api_key_dict)


Provider Code Outside llms

This route and its logging dispatch add TypeSafe-specific behavior outside llms/, violating the repository isolation requirement that must be satisfied before merge

Rule Used: What: Avoid writing provider-specific code outside of the llms/ directory. Why: This practice ensures better maintainability and reduces complexity over time. Good: ```python # Handle provider-specific logic within llms/vertex_ai/transformation.py ... (source)

Note: If this suggestion doesn't match your team's coding style, reply to this and let me know. I'll remember it for next time!

TypeSafe pass-through routes intentionally use the shared proxy passthrough layer; moving dispatch into llms would break established provider passthrough architecture.

That’s fair. The route delegates to the shared passthrough machinery, and the TypeSafe-specific logging handler is placed alongside the existing provider passthrough handlers under proxy/pass_through_endpoints/llm_provider_handlers. This is proxy integration glue rather than provider implementation that belongs in llms/; moving it there would break the established passthrough architecture. I’m withdrawing this finding.

Shared passthrough routing and success handlers intentionally live under proxy; provider-specific completion adapters cannot replace this endpoint forwarding and spend-log integration.

codecov · 2026-09-17T16:09:27Z

Codecov Report

❌ Patch coverage is 93.15068% with 5 lines in your changes missing coverage. Please review.

Files with missing lines	Patch %	Lines
...r_handlers/typesafe_passthrough_logging_handler.py	90.90%	5 Missing ⚠️

📢 Thoughts on this report? Let us know!

Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>

mateo-berri · 2026-09-17T17:33:22Z

@greptileai

mateo-berri · 2026-09-17T17:33:42Z

@greptileai

mateo-berri · 2026-09-17T17:44:18Z

@greptileai

mateo-berri · 2026-09-17T17:50:16Z

bugbot run

cursor

Stale Bugbot comment from a previous run.

Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>

mateo-berri · 2026-09-17T18:18:04Z

@greptileai

greptile-apps · 2026-09-17T18:21:32Z

+    updated_url: Final = base_url.copy_with(
+        path=HttpPassThroughEndpointHelpers.join_base_and_endpoint_path(base_url, normalized_endpoint),
+    )


POST Query Parameters Dropped

For TypeSafe JSON POST requests such as /typesafe/v1/systemone?trace=yes, this URL construction no longer includes the incoming query string. The shared JSON passthrough path only reads query parameters from the request body's query_params field, so normal URL query parameters are omitted from the upstream request. The updated test hides this behavior by merging request.query_params inside its fake endpoint rather than exercising the production forwarding path.

The shared passthrough forwards request.query_params itself (pass_through_endpoints.py line 994, then params= on the JSON POST build_request), so removing them from the target avoids duplication.

You're right. The shared JSON passthrough adds request.query_params as the params= argument when building the upstream request, so keeping the query string out of target avoids duplication while still forwarding it. The test's copy_merge_params() models that shared behavior; it is not masking a production omission. This comment is incorrect and can be dismissed.

mateo-berri · 2026-09-17T18:24:01Z

bugbot run

cursor

Stale Bugbot comment from a previous run.

mateo-berri · 2026-09-17T18:31:16Z

@greptileai

Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>

mateo-berri · 2026-09-17T21:20:31Z

@greptileai

mateo-berri · 2026-09-17T22:05:31Z

bugbot run

cursor

Stale Bugbot comment from a previous run.

mateo-berri · 2026-09-17T22:44:25Z

@greptileai

mateo-berri · 2026-09-17T22:53:20Z

bugbot run

cursor

✅ Bugbot reviewed your changes and found no new issues!

Comment @cursor review or bugbot run to trigger another review on this PR

^{Reviewed by Cursor Bugbot for commit 1feaa48. Configure here.}

mateo-berri

LGTM

BerriAI#41607 registered the typesafe pass-through with a route that only accepted GET and POST, so a PUT, DELETE or PATCH to /typesafe/... came back 405 before reaching the upstream. CircleCI's pass-through method test caught it, but that lane does not run on the PR gate, so the mapped unit test now covers the same invariant for typesafe The same CircleCI run also failed test_models_by_provider because typesafe is not a key of models_by_provider. Registering it there would satisfy the assertion without changing behaviour: typesafe has no LlmProviders member, so a typesafe/* deployment never loads and get_valid_models returns nothing, and its spend is priced straight from model_cost. The test already skips search-mode providers for that reason, so it now skips evaluation mode too

Partial backport of upstream BerriAI#41607 (price rows and the "evaluation" model mode only; the /typesafe pass-through is not carried), plus the test_models_by_provider skip from BerriAI#41723. The Jev classifier (BerriAI#41615) prices its calls from typesafe/jev-* registry rows, so they must exist even when LITELLM_LOCAL_MODEL_COST_MAP is set. Drop this commit when rebasing onto a release that contains BerriAI#41607. Co-Authored-By: Mateo Wang <277851410+mateo-berri@users.noreply.github.com> Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>

feat(typesafe): backport #41607 to stable/1.101.x for v1.101.1

feat(typesafe): backport #41607 to stable/1.100.x for v1.100.2

feat(typesafe): backport #41607 to stable/1.99.x for v1.99.2

…acking Backport of BerriAI#41607 to stable/1.102.x. Cherry-picked from deb9d8a (main). Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>

…acking Backport of BerriAI#41607 to stable/1.99.x. Cherry-picked from deb9d8a (main). Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>

mateo-berri and others added 2 commits September 17, 2026 15:53

feat(proxy): add TypeSafe Jev passthrough spend tracking

2dc9697

Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>

refactor(proxy): simplify TypeSafe passthrough pricing lookup and route

78eb92c

Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>

devin-ai-integration Bot requested a review from a team September 17, 2026 15:57

devin-ai-integration Bot assigned mateo-berri Sep 17, 2026

devin-ai-integration Bot requested review from kerry-berri, mateo-berri and ryan-crabbe-berri as code owners September 17, 2026 15:57

fix(proxy): satisfy TypeSafe CI gates

9470aa4

Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>

greptile-apps Bot reviewed Sep 17, 2026

View reviewed changes

mateo-berri and others added 2 commits September 17, 2026 16:20

fix(proxy): satisfy TypeSafe CI gates

7fca7fa

Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>

fix(proxy): expose TypeSafe passthrough on gateway

4f3b90b

Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>

This was referenced Sep 17, 2026

docs: add TypeSafe AI Jev pass-through page BerriAI/litellm-docs#1524

Merged

feat(router): add TypeSafe Jev as a complexity router classifier #41615

Merged

cursor Bot reviewed Sep 17, 2026

View reviewed changes

Comment thread litellm/proxy/pass_through_endpoints/llm_passthrough_endpoints.py Outdated

fix(proxy): stop duplicating query params on the TypeSafe passthrough

b2ef8da

Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>

greptile-apps Bot reviewed Sep 17, 2026

View reviewed changes

cursor Bot reviewed Sep 17, 2026

View reviewed changes

shijiew mentioned this pull request Sep 17, 2026

[Feature]: Jev router — use TypeSafe Jev (System One decision model) as a routing classifier with CEL jev.* variables maximhq/bifrost#7278

Open

1 task done

chore(proxy): merge origin/main into TypeSafe passthrough

dcbb773

Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>

cursor Bot reviewed Sep 17, 2026

View reviewed changes

Comment thread ...m/proxy/pass_through_endpoints/llm_provider_handlers/typesafe_passthrough_logging_handler.py

fix(proxy): log TypeSafe calls that name no model as unknown

1feaa48

cursor Bot reviewed Sep 17, 2026

View reviewed changes

mateo-berri approved these changes Sep 17, 2026

View reviewed changes

mateo-berri merged commit deb9d8a into main Sep 17, 2026
89 of 90 checks passed

mateo-berri deleted the litellm_typesafe_passthrough branch September 17, 2026 23:37

mateo-berri mentioned this pull request Sep 18, 2026

[Bug]: POST /model/new with a typesafe/* model saves the row then answers 500 because typesafe is not a routable provider #41714

Open

1 task done

yuneng-berri mentioned this pull request Sep 18, 2026

fix(proxy): forward every method on the typesafe pass-through route #41723

Merged

4 of 6 tasks

This was referenced Sep 18, 2026

fix(proxy): enforce virtual key budgets for JEV test routing #41879

Merged

feat(auto-router): add JEV classifier alongside LLM classifier #41886

Merged

dragon3 mentioned this pull request Sep 19, 2026

Support TypeSafe Jev (System One) as a backend API schema and endpoint theagentrouter/agent-router#2712

Closed

srinivas-jay mentioned this pull request Sep 19, 2026

feat: upgrade LiteLLM to 1.101.0 with TypeSafe pass-through aialchemylabs/aialchemy-llm-gateway#17

Merged

n24q02m mentioned this pull request Sep 20, 2026

Jev classifier: confidence-gated route-down + fail-expensive option #42139

Open

mateo-berri added a commit that referenced this pull request Sep 23, 2026

Merge pull request #42596 from BerriAI/litellm_cherrypick_1_101_x

432215e

feat(typesafe): backport #41607 to stable/1.101.x for v1.101.1

mateo-berri added a commit that referenced this pull request Sep 23, 2026

Merge pull request #42597 from BerriAI/litellm_cherrypick_1_100_x

9c1216a

feat(typesafe): backport #41607 to stable/1.100.x for v1.100.2

mateo-berri added a commit that referenced this pull request Sep 23, 2026

Merge pull request #42598 from BerriAI/litellm_cherrypick_1_99_x

1394d33

feat(typesafe): backport #41607 to stable/1.99.x for v1.99.2

This was referenced Sep 23, 2026

chore(backport): add the remaining jev changes to stable/1.100.x #42668

Open

chore(backport): add the remaining jev changes to stable/1.99.x #42669

Open

Conversation

devin-ai-integration Bot commented Sep 17, 2026 • edited by mateo-berri Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

TLDR

User Flow

Relevant issues

Affected release

Linear ticket

Pre-Submission checklist

Delays in PR merge?

Screenshots / Proof of Fix

Before (87650bf)

After (1feaa48)

Type

Caveats (if any)

Low

Final Attestation

Uh oh!

devin-ai-integration Bot commented Sep 17, 2026

🤖 Devin AI Engineer

Uh oh!

codspeed Bot commented Sep 17, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Merging this PR will not alter performance

Uh oh!

greptile-apps Bot commented Sep 17, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Confidence Score: 5/5

Findings

Summary

Uh oh!

greptile-apps Bot Sep 17, 2026

Choose a reason for hiding this comment

Uh oh!

greptile-apps Bot Sep 17, 2026

Choose a reason for hiding this comment

Uh oh!

mateo-berri Sep 17, 2026

Choose a reason for hiding this comment

Uh oh!

greptile-apps Bot Sep 17, 2026

Choose a reason for hiding this comment

Uh oh!

mateo-berri Sep 17, 2026

Choose a reason for hiding this comment

Uh oh!

codecov Bot commented Sep 17, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Codecov Report

Uh oh!

mateo-berri commented Sep 17, 2026

Uh oh!

mateo-berri commented Sep 17, 2026

Uh oh!

mateo-berri commented Sep 17, 2026

Uh oh!

mateo-berri commented Sep 17, 2026

Uh oh!

cursor Bot left a comment • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

mateo-berri commented Sep 17, 2026

Uh oh!

greptile-apps Bot Sep 17, 2026

Choose a reason for hiding this comment

Uh oh!

mateo-berri Sep 17, 2026

Choose a reason for hiding this comment

Uh oh!

greptile-apps Bot Sep 17, 2026

Choose a reason for hiding this comment

Uh oh!

mateo-berri commented Sep 17, 2026

Uh oh!

cursor Bot left a comment • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

devin-ai-integration Bot commented Sep 17, 2026 •

edited by mateo-berri

Loading

Before (`87650bf`)

After (`1feaa48`)

codspeed Bot commented Sep 17, 2026 •

edited

Loading

greptile-apps Bot commented Sep 17, 2026 •

edited

Loading

codecov Bot commented Sep 17, 2026 •

edited

Loading

cursor Bot left a comment •

edited

Loading

cursor Bot left a comment •

edited

Loading

cursor Bot left a comment •

edited

Loading