Skip to content

feat(pi,omp): add runtime model fallback #2358

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
shunkakinoki merged 7 commits into from
Aug 12, 2026
Merged

feat(pi,omp): add runtime model fallback #2358

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
7 commits
Select commit Hold shift + click to select a range
7e57181
feat(pi,omp): add runtime model fallback with shared policy
shunkakinoki Aug 12, 2026
88eca69
fix(fallback): expand json arrays for treefmt and skip unselectable m…
shunkakinoki Aug 12, 2026
fface40
fix(fallback): validate config, honor manual model switches, scope sp…
shunkakinoki Aug 12, 2026
1def52d
refactor(fallback): use OMP native retry chains, inline pi policy
shunkakinoki Aug 12, 2026
1a4d715
fix(omp): give GPT-pinned roles their own fallback chains
shunkakinoki Aug 12, 2026
518f1e1
test(omp): guard fallback chain tails by base selector too
shunkakinoki Aug 12, 2026
5000be0
feat(pi): use free-tier CLIProxy fallback chain
shunkakinoki Aug 12, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
34 changes: 34 additions & 0 deletions config/omp/config.tpl.yml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -299,6 +299,40 @@ retry:

# Base delay between retries in milliseconds. Upstream default: 2000
baseDelayMs: 2000
# modelFallback

# Let retry recovery switch to the chains below. Upstream default: true
modelFallback: true
# fallbackChains

# Ordered alternates per chain key. The `default` chain is expanded to every
# model role without its own chain, so role turns fall back too.
#
# Candidates are the entries AFTER the current model in its chain, so a role
# whose model sits last in the chain it inherits has no candidates at all.
# The roles pinned to __GPT__ therefore get their own chain rather than
# inheriting `default`, which ends on __GPT__.
fallbackChains:
default:
- "openai-codex/__GPT_LUNA__"
- "openai-codex/__GPT_CODEX_SPARK__"
- "openai-codex/__GPT__"
slow:
- "openai-codex/__GPT_CODEX__"
- "openai-codex/__GPT_LUNA__"
- "cliproxyapi/__DEEPSEEK_FLASH__"
vision:
- "openai-codex/__GPT_CODEX__"
- "openai-codex/__GPT_LUNA__"
- "cliproxyapi/__DEEPSEEK_FLASH__"
plan:
- "openai-codex/__GPT_CODEX__"
- "openai-codex/__GPT_LUNA__"
- "cliproxyapi/__DEEPSEEK_FLASH__"
# fallbackRevertPolicy

# Return to the primary once its cooldown expires. Upstream default: cooldown-expiry
fallbackRevertPolicy: "cooldown-expiry"
# --- Compaction --------------------------------------------------------------
compaction:
# enabled
Expand Down
34 changes: 34 additions & 0 deletions config/omp/config.yml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -299,6 +299,40 @@ retry:

# Base delay between retries in milliseconds. Upstream default: 2000
baseDelayMs: 2000
# modelFallback

# Let retry recovery switch to the chains below. Upstream default: true
modelFallback: true
# fallbackChains

# Ordered alternates per chain key. The `default` chain is expanded to every
# model role without its own chain, so role turns fall back too.
#
# Candidates are the entries AFTER the current model in its chain, so a role
# whose model sits last in the chain it inherits has no candidates at all.
# The roles pinned to gpt-5.6-sol therefore get their own chain rather than
# inheriting `default`, which ends on gpt-5.6-sol.
fallbackChains:
default:
- "openai-codex/gpt-5.6-luna"
- "openai-codex/gpt-5.3-codex-spark"
- "openai-codex/gpt-5.6-sol"
slow:
- "openai-codex/gpt-5.3-codex"
- "openai-codex/gpt-5.6-luna"
- "cliproxyapi/deepseek-v4-flash"
vision:
- "openai-codex/gpt-5.3-codex"
- "openai-codex/gpt-5.6-luna"
- "cliproxyapi/deepseek-v4-flash"
plan:
- "openai-codex/gpt-5.3-codex"
- "openai-codex/gpt-5.6-luna"
- "cliproxyapi/deepseek-v4-flash"
# fallbackRevertPolicy

# Return to the primary once its cooldown expires. Upstream default: cooldown-expiry
fallbackRevertPolicy: "cooldown-expiry"
# --- Compaction --------------------------------------------------------------
compaction:
# enabled
Expand Down
8 changes: 8 additions & 0 deletions config/pi/default.nix
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -15,4 +15,12 @@ _: {
source = ./moshi-hooks.ts;
force = true;
};
home.file.".pi/agent/fallback.json" = {
source = ./fallback.json;
force = true;
};
home.file.".pi/agent/extensions/fallback.ts" = {
source = ./fallback.ts;
force = true;
};
}
23 changes: 23 additions & 0 deletions config/pi/fallback.json
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,23 @@
{
"enabled": true,
"retry_on_errors": [
401,
404,
429,
500,
502,
503,
504
],
"retryable_error_patterns": [
"unknown provider for model"
],
"max_fallback_attempts": 5,
"cooldown_seconds": 60,
"notify_on_fallback": true,
"fallback_models": [
"cliproxyapi/gemma-4-31b-it",
"cliproxyapi/glm-4.7",
"cliproxyapi/free"
]
}
23 changes: 23 additions & 0 deletions config/pi/fallback.tpl.json
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,23 @@
{
"enabled": true,
"retry_on_errors": [
401,
404,
429,
500,
502,
503,
504
],
"retryable_error_patterns": [
"unknown provider for model"
],
"max_fallback_attempts": 5,
"cooldown_seconds": 60,
"notify_on_fallback": true,
"fallback_models": [
"cliproxyapi/__GEMMA__",
"cliproxyapi/__GLM__",
"cliproxyapi/free"
]
}
Loading
Loading