Commit
This commit does not belong to any branch on this repository, and may belong to a fork outside of the repository.
feat: aiproxy modelinfo and dashboard and model rpm limit (#5291)
* feat: model info * fix: model config vision * feat: aiproxy dashboard api * fix: two week and pg hour format * fix: model tag name * feat: model rpm limit * fix: ci * feat: search log with code type * feat: resp detail buf use pool * feat: no need init client, use ctx * fix: lint * feat: admin api log filed * feat: log usage * feat: auto retry * fix: retry channel exhausted, use first channel * feat: init monitor * feat: auto ban error rate and auto test unban * fix: getChannelWithFallback * feat: support google thinking * fix: monitor * feat: get log detail * feat: no need channel config * feat: key validate * feat: add model error auto ban optioon * feat: gemini tool * feat: gemini openai sdk * fix: option keys * feat: do not save access at * fix: del no use options * fix: del no use options * fix: auto test banned models need return when get from redis error happend * fix: remove channel db hook * chore: clean detail only after insert it * fix: err print on debug * fix: cache update * feat: group consume level rpm ratio * fix: error return * feat: decode svg * fix: check is image * fix: reply raw 429 message * feat: req and resp body max size limit * fix: _ import lint * fix: get token encoder log * fix: sum used amount * fix: delete no need cache * feat: dashboard rpm * feat: dashboard tpm * feat: step modelinfo * feat: yi * fix: yi * feat: debug banned * chore: bump go mod * chore: bump go mod * fix: save model time parse * feat: fill dash carts gaps * feat: fill dash carts gaps * chore: go mod tidy * feat: dashboard timespan * feat: dashboard timespan from query * feat: decouple request paths * feat: group model tmp limit * feat: decoupling url paths * fix: check balance * refactor: relay handler * refactor: post relay * feat: fill gaps before and after point * fix: qwen long tokens * feat: get rpm from redis * fix: fill gaps * fix: log error * fix: token not fount err log * fix: if err resp is not json, replay raw content * fix: do not save same response body and content * fix: save resp json or empty * feat: sort distinct values * fix: token models * feat: redis clean expired cache * feat: atomic model cache * feat: consume * feat: group custom model rpm tpm * fix: models * fix: v1 route * fix: cros * feat: rate limit err log record * fix: rpush * fix: dashboard time span * feat: group model list adjusted tpm rpm * feat: baichuan model config * fix: rpm limit recore ignore empty channel id * feat: disable model config * feat: internal token * fix: lint * fix: recore req to redis * feat: option from env * fix: internal token option key * fix: ignore redis ping error * fix: ignore redis ping error * fix: subscription * fix: subscription * feat: precheck group balance * fix: consume nil pointer * feat: log balance * feat: ip log * fix: group disable * fix: non stream context cancel * feat: amount log * fix: balance and amount log format * fix: do not skip empty * fix: reason system prompt * feat: doubao and moonshot model * feat: disable model config can load existed model * chore: add shutdown timeout duration to 600 sec * feat: dashboard data build whit concurrent * feat: logs data build whit concurrent * fix: monitor remove banned model * feat: split think * fix: skip enpty think * fix: do not store large resp * fix: reat limit script * fix: reat limit use micro second * fix: ignore gemini input count error * feat: calude model config * fix: claude stream usage resp * fix: claude stream usage resp * fix: claude stream usage resp * feat: auto create sqlite dir * feat: log detail body truncated * chore: add body conv commend * feat: monitor ignore error rate compute when is success request * feat: ollama usage support * feat: baseurl embed v1 prefix * feat: limit detail record size * feat: split think config * feat: channel default priority * fix: rate limit message * feat: channel meta api * feat: add channel key validate help message * fix: channel config update * fix: split think * fix: claude api * fix: record total tokens * chore: bump go mod * chore: bump go mod * feat: qwen open source vl models * fix: qwen2.5 vl tool choice * feat: stt audio duration * feat: ali paraformer price * fix: stt usage * feat: qwen mt * fix: render when split skip * feat: sealos realname check * feat: gemini usage support * fix: lint * fix: error message * fix: lint * fix: search token * fix: no real name limit han message * feat: gemini model config * fix: get group error hans message * fix: get group dashboard models * feat: channel and token model search * feat: support ali completions * feat: internal group and search optimize * feat: conv gemini tool choice * fix: gemini empty tool parameters * chore: env readme * fix: ci lint
- Loading branch information