[backryun]

Summary

Modernize the Arena web-session provider (formerly LMArena; wire id stays lmarena) for arena.ai.

• Chat via https://arena.ai/nextjs-api/stream/create-evaluation with Chrome TLS impersonation (tls-client-node) and optional browser-minted recaptchaV3Token.
• Static Direct-chat catalog (2026-07-09 scrape, TypeScript only — no scrape JSON in-repo):
  • Text + Search (48) → chat registry
  • Image (27) → IMAGE_PROVIDERS only (not the chat provider page)
• Disable live arena.ai HTML model discovery; resolve public names → Arena UUIDs from the static seed.
• Soft-exclude models that 404/502; slow/stop bulk test-all probes for this provider.
• Specialty catalog: do not fold image/video rows into chat specialty when a chat registry exists (lmarena / openai / xai).
• UI rebrand: display name Arena (Free); theme-aware icons arena-light.svg / arena-dark.svg.
• Keep split Supabase SSR cookie reconstruction for arena-auth-prod-v1.*.

Wire / back-compat

Field	Value
Provider id	lmarena (unchanged)
Alias	lma
Display name	Arena (Free)
Website	https://arena.ai

Validation

• Targeted unit tests (lmarena executor, split-cookie, specialty catalog, themed icons, web credentials)
• Live smoke with real session cookie (operator-verified)
• CI gates on this PR (lint / unit / quality)

Local examples:

node --import tsx/esm --test \
  tests/unit/lmarena-provider.test.ts \
  tests/unit/lmarena-split-cookie-4271.test.ts \
  tests/unit/provider-icon-arena-themed.test.ts \
  tests/unit/bailian-coding-plan-provider.test.ts

Tests added / updated

• tests/unit/lmarena-provider.test.ts
• tests/unit/lmarena-split-cookie-4271.test.ts
• tests/unit/provider-icon-arena-themed.test.ts
• tests/unit/bailian-coding-plan-provider.test.ts (specialty no image dump)
• related web-cookie / unsupported-validation / catalog tests

Reviewer notes

• Image models are listed for the image catalog only; full /v1/images/generations path for Arena is out of scope unless follow-up.
• Cloudflare may still require a residential path and/or operator-supplied reCAPTCHA v3 token in hard environments.
• Scrape JSON stays local/desktop only; allowlist ships as directModels.ts.

[gemini-code-assist]

Code Review

This pull request updates the LMArena provider integration to support the new direct-battle evaluation endpoint and handles the updated AI-SDK SSE stream prefixes. It introduces dynamic model discovery for LMArena by fetching and parsing the initial models directly from arena.ai, complete with caching and display-name-to-internal-ID mapping. Additionally, it refactors the UI validation flow to treat "unsupported" validation results as non-blocking warnings (rendered as a neutral "N/A" badge) rather than blocking errors, and updates LMArena credential hints to reflect Supabase SSR chunked cookies. The reviewer provided several constructive feedback items for open-sse/executors/lmarena.ts, including guarding against null/undefined request bodies in transformRequest, caching the active fetch promise in getLMArenaModels to prevent thundering herd issues, optimizing string trimming and redundant operations in SSE parsing, and adding a safety guard in parseLMArenaInitialModels to prevent slicing errors.

Important

The consumer version of Gemini Code Assist on GitHub is being sunset. Starting June 18, 2026, new organization installations will be blocked, and all code review activity will officially cease on July 17, 2026.
For more details on the timeline and next steps, please review the Help Documentation.

[diegosouzapw]

🔀 Cycle moved: base retargeted to release/v3.8.46 — release/v3.8.45 is frozen for the release in progress (#6313). No action needed.

[backryun]

@diegosouzapw I’d like to request a quick merge. It’s been five days, and I’ve done my best to ensure there are no conflicts with recent commits.

[diegosouzapw]

@backryun thanks for the Arena web provider modernization — merged into release/v3.8.47 (fixture + fallback-suite aligned to the new registry entry). 🙏