v7.2.150
Linux release assets
CLIProxyAPI_<version>_linux_<arch>.tar.gzis the default Linux build. It supports dynamic library plugins and is built against a GLIBC 2.17 baseline.CLIProxyAPI_<version>_linux_<arch>_no-plugin.tar.gzis the portable Linux build for musl-based or older systems such as OpenWrt. It does not support dynamic library plugins.
FreeBSD release assets
CLIProxyAPI_<version>_freebsd_aarch64_no-plugin.tar.gzis the FreeBSD arm64 build. It is built without CGO and does not support dynamic library plugins.
Changelog
- fix(openai-compat): honor bounded rate-limit waits (6f16121)
- fix(translator/gemini): nest image parts inside functionResponse (728ea8b)
- fix(translator/claude): defer message_delta and cache streaming usage (f804fb5)
- perf(antigravity): batch replay degradation rewrites (#5461) (e44432a)
- fix(translator/claude): handle incomplete status and terminal state on max_tokens (ba2cdea)
- feat(plugin): omit stream chunk history on payload chunks for schema v5 (649a8bd)
- Merge pull request #5452 from deathemperor/docs/add-infinitus (b9110ec)
- Merge pull request #5435 from huangruiteng/codex/openai-compat-bounded-rate-limit-waits (9ec2bc2)
- fix(translator/interactions): avoid tool name collisions with Antigravity intrinsic tools (6a26e92)
- fix(translator/claude): downgrade strict mode when schema misses required properties (aa36527)
- fix(auth): preserve concurrent modifications during auth refresh and preparation (4c1bebe)
- fix(antigravity): tighten replacement offset guard in tool provenance degradation (1c45093)
- feat(claude): implement 2.1.258 billing header fingerprint chain and upstream request continuity (086ad91)
- feat(claude): add 2.1.258 dynamic beta headers, model fallbacks, and paired cache TTL (d7052c9)
- feat(claude): add Fable 5.1 reporting outcomes block and post-payload reconciliation (de4aa60)
- feat(claude): harden probe and helper request classification, diagnostics isolation, and late cloaking (4a5ab53)
- fix(translator): preserve Gemini prompt cache by demoting mid-session developer messages (#5490) (0fe19ed)
- fix(translator): flush pending developer notice before intervening user turn (e56fae8)
- fix(antigravity): use ContentHasGeminiFunctionResponse instead of gjson projection (f2041a2)
- fix(gemini): ensure functionResponse normalizes to user role in Gemini request normalizer (f6d19a3)
- perf(antigravity): batch reasoning replay mutations (acf919c)
- feat(models): add gpt-6-astra model and update codex client configurations (c77b136)
What's Changed
- perf(antigravity): batch replay degradation rewrites by @rome-xi in #5461
- docs: add Infinitus to "Who is with us?" by @deathemperor in #5452
- fix(openai-compat): honor bounded rate-limit waits by @huangruiteng in #5435
- feat(claude): upgrade Claude Code 2.1.258 wire fingerprint and model behavioral parity by @sususu98 in #5456
- fix(translator): preserve Gemini prompt cache by demoting mid-session developer messages (#5490) by @sususu98 in #5496
- perf(antigravity): batch reasoning replay mutations by @rome-xi in #5498
New Contributors
- @rome-xi made their first contribution in #5461
- @deathemperor made their first contribution in #5452
- @huangruiteng made their first contribution in #5435
Full Changelog: v7.2.149...v7.2.150