{"schema_version":"0.1","id":"33d81162-7b09-470b-a65d-cd0884394220","display_name":"Li Wei","handle":"li-wei-evals","avatar_url":null,"positioning":"Chinese-language model reviews, run on my own machines first.","topics":["Model reviews","Chinese models","Agents"],"region":"Shanghai, China","languages":{"content":["Chinese"],"communication":["Chinese","English"]},"time_zone":"Asia/Shanghai","accounts":[{"platform":"bilibili","handle":"li-wei-evals","url":null,"ownership_status":"self_reported","metric_snapshot":{"metric":"followers","value":86300,"unit":null,"period":null,"sample_size":null,"as_of":"2026-11-03T09:05:30.284Z","source_type":"self_reported","historical":false},"evidence":[]},{"platform":"x","handle":"li-wei-evals","url":null,"ownership_status":"self_reported","metric_snapshot":{"metric":"followers","value":3100,"unit":null,"period":null,"sample_size":null,"as_of":"2026-11-03T12:05:30.284Z","source_type":"self_reported","historical":false},"evidence":[]}],"works":[{"id":"4c5c0d2b-fe6c-47f7-94a2-f610bddff5a6","type":"video","title":"Qwen3.5-35B-A3B 本地实测：值不值得换","role":null,"personal_contribution":null,"outcome":null,"limitations":null,"version":null,"links":[],"disclosure":"independent","pinned":true,"run_id":null,"evidence":[{"type":"source_linked","label":"Source linked","scope":"The link points to this creator's published work.","does_not_establish":"Does not establish the numbers in it.","review_status":"pending","version":null,"date":"2026-11-01T12:05:30.284Z"}],"updated_at":"2026-11-01T12:05:30.284Z","date":"2026-11-01","visibility":"public"},{"id":"d80af1dd-ff1c-4be2-a0ea-b5edad7700db","type":"run","title":"gemma-3-12b-it on suite-standard@1.0","role":"Ran the suite","personal_contribution":"Ran suite-standard@1.0 on gemma3:12b-it-q4_K_M, NVIDIA GeForce RTX 4090 · 24 GB.","outcome":"code-intervals 2/3 · tool-flight not supported · needle-8k 3/3 · 89.3 tok/s decode at speed@0","limitations":"Applies to this build, machine and suite version. A single community run on the submitter's machine, not part of headline numbers.","version":"suite-standard@1.0 · Q4_K_M","links":[],"disclosure":"not_stated","pinned":false,"run_id":"46083bcc-f543-4f5d-accf-8d002e277469","evidence":[{"type":"verified","label":"Verified","scope":"1 of 2 tasks regraded from the event log; code tasks not regraded","does_not_establish":"Does not establish reliability on other machines or that the platform executed the run.","review_status":"passed","version":"suite-standard@1.0 · Q4_K_M","date":"2026-11-06T10:53:30.284Z"}],"updated_at":"2026-11-06T10:53:30.284Z","date":"2026-11-06","visibility":"public"},{"id":"4032f284-af4c-4e5f-bb0c-95cff9697376","type":"run","title":"MiniCPM5-2B on suite-standard@1.0","role":"Ran the suite","personal_contribution":"Queued suite-standard@1.0 for openbmb/minicpm5-2b:q4_K_M on BenchClub cloud, NVIDIA L40S · 48 GB, pinned environment.","outcome":"code-intervals 0/3 · tool-flight 1/3 · needle-8k 3/3 · 297.2 tok/s decode at speed@0","limitations":"Applies to this build and suite version on BenchClub's pinned cloud environment. Counts toward headline results.","version":"suite-standard@1.0 · Q4_K_M","links":[],"disclosure":"not_stated","pinned":false,"run_id":"0369b59d-cf8a-40e0-8264-601e9b72e243","evidence":[{"type":"verified","label":"Verified","scope":"Run on BenchClub cloud; all 3 tasks graded by the platform","does_not_establish":"Does not establish results on other hardware.","review_status":"passed","version":"suite-standard@1.0 · Q4_K_M","date":"2026-11-06T06:26:18.678Z"}],"updated_at":"2026-11-06T06:26:18.678Z","date":"2026-11-06","visibility":"public"},{"id":"59ccd319-fef0-48d1-bf15-6777740359f2","type":"run","title":"Mistral-7B-Instruct-v0.2 on suite-standard@1.0","role":"Ran the suite","personal_contribution":"Queued suite-standard@1.0 for mistral:7b-instruct-v0.2-q4_K_M on BenchClub cloud, NVIDIA L40S · 48 GB, pinned environment.","outcome":"code-intervals 1/3 · tool-flight not supported · needle-8k 2/3 · 136.1 tok/s decode at speed@0","limitations":"Applies to this build and suite version on BenchClub's pinned cloud environment. Counts toward headline results.","version":"suite-standard@1.0 · Q4_K_M","links":[],"disclosure":"not_stated","pinned":false,"run_id":"e4bb5154-8188-4f6f-ac2e-8367f6a6f399","evidence":[{"type":"verified","label":"Verified","scope":"Run on BenchClub cloud; all 2 tasks graded by the platform","does_not_establish":"Does not establish results on other hardware.","review_status":"passed","version":"suite-standard@1.0 · Q4_K_M","date":"2026-11-02T17:13:16.305Z"}],"updated_at":"2026-11-02T17:13:16.305Z","date":"2026-11-02","visibility":"public"},{"id":"04af95b3-3001-4d15-ac00-e6b4841085de","type":"run","title":"gemma-3-27b-it on suite-standard@1.0","role":"Ran the suite","personal_contribution":"Ran suite-standard@1.0 on gemma3:27b-it-q4_K_M, Apple M4 Max · 128 GB.","outcome":"code-intervals 2/3 · tool-flight not supported · needle-8k 2/3 · 19.7 tok/s decode at speed@0","limitations":"Applies to this build, machine and suite version. A single community run on the submitter's machine, not part of headline numbers.","version":"suite-standard@1.0 · Q4_K_M","links":[],"disclosure":"not_stated","pinned":false,"run_id":"c34b3505-13f6-4380-87b1-aa2ef6df33b4","evidence":[{"type":"verified","label":"Verified","scope":"1 of 2 tasks regraded from the event log; code tasks not regraded","does_not_establish":"Does not establish reliability on other machines or that the platform executed the run.","review_status":"passed","version":"suite-standard@1.0 · Q4_K_M","date":"2026-11-01T05:59:01.793Z"}],"updated_at":"2026-11-01T05:59:01.793Z","date":"2026-11-01","visibility":"public"},{"id":"da81ae74-7f15-46fb-a277-a2722e1a9663","type":"run","title":"Qwen3-8B on suite-standard@1.0","role":"Ran the suite","personal_contribution":"Queued suite-standard@1.0 for qwen3:8b-q4_K_M on BenchClub cloud, NVIDIA L40S · 48 GB, pinned environment.","outcome":"code-intervals 3/3 · tool-flight 2/3 · needle-8k 3/3 · 119.1 tok/s decode at speed@0","limitations":"Applies to this build and suite version on BenchClub's pinned cloud environment. Counts toward headline results.","version":"suite-standard@1.0 · Q4_K_M","links":[],"disclosure":"not_stated","pinned":false,"run_id":"aba0337b-5321-4558-b038-3617a1e2cce7","evidence":[{"type":"verified","label":"Verified","scope":"Run on BenchClub cloud; all 3 tasks graded by the platform","does_not_establish":"Does not establish results on other hardware.","review_status":"passed","version":"suite-standard@1.0 · Q4_K_M","date":"2026-10-29T17:45:58.059Z"}],"updated_at":"2026-10-29T17:45:58.059Z","date":"2026-10-29","visibility":"public"},{"id":"8528bbca-a825-42d5-95d1-b9c145204ed0","type":"run","title":"Mistral-7B-Instruct-v0.2 on suite-standard@1.0","role":"Ran the suite","personal_contribution":"Ran suite-standard@1.0 on mistral:7b-instruct-v0.2-q4_K_M, NVIDIA GeForce RTX 4090 · 24 GB.","outcome":"code-intervals 0/3 · tool-flight not supported · needle-8k 3/3 · 156.2 tok/s decode at speed@0","limitations":"Applies to this build, machine and suite version. A single community run on the submitter's machine, not part of headline numbers.","version":"suite-standard@1.0 · Q4_K_M","links":[],"disclosure":"not_stated","pinned":false,"run_id":"0c70ab3e-afb2-46c8-b6f4-90c8f0075f41","evidence":[{"type":"verified","label":"Verified","scope":"1 of 2 tasks regraded from the event log; code tasks not regraded","does_not_establish":"Does not establish reliability on other machines or that the platform executed the run.","review_status":"passed","version":"suite-standard@1.0 · Q4_K_M","date":"2026-10-29T12:14:47.418Z"}],"updated_at":"2026-10-29T12:14:47.418Z","date":"2026-10-29","visibility":"public"},{"id":"17d36869-4186-45bc-b332-9751ca0d4bd5","type":"run","title":"Qwen3-14B on suite-standard@1.0","role":"Ran the suite","personal_contribution":"Queued suite-standard@1.0 for qwen3:14b-q4_K_M on BenchClub cloud, NVIDIA L40S · 48 GB, pinned environment.","outcome":"code-intervals 3/3 · tool-flight 2/3 · needle-8k 3/3 · 67.1 tok/s decode at speed@0","limitations":"Applies to this build and suite version on BenchClub's pinned cloud environment. Counts toward headline results.","version":"suite-standard@1.0 · Q4_K_M","links":[],"disclosure":"not_stated","pinned":false,"run_id":"abf900ce-08be-44f6-904b-34391a000168","evidence":[{"type":"verified","label":"Verified","scope":"Run on BenchClub cloud; all 3 tasks graded by the platform","does_not_establish":"Does not establish results on other hardware.","review_status":"passed","version":"suite-standard@1.0 · Q4_K_M","date":"2026-10-24T17:04:07.194Z"}],"updated_at":"2026-10-24T17:04:07.194Z","date":"2026-10-24","visibility":"public"},{"id":"e4278c04-cd3b-4814-8bf1-e02c8611e35b","type":"run","title":"Llama-3.1-8B-Instruct on suite-standard@1.0","role":"Ran the suite","personal_contribution":"Ran suite-standard@1.0 on llama3.1:8b-instruct-q4_K_M, Apple M4 Max · 128 GB.","outcome":"code-intervals 2/3 · tool-flight 1/3 · needle-8k 1/3 · 67.8 tok/s decode at speed@0","limitations":"Applies to this build, machine and suite version. A single community run on the submitter's machine, not part of headline numbers.","version":"suite-standard@1.0 · Q4_K_M","links":[],"disclosure":"not_stated","pinned":false,"run_id":"229dd708-6822-4df1-8439-468da1b64fc3","evidence":[{"type":"verified","label":"Verified","scope":"2 of 3 tasks regraded from the event log; code tasks not regraded","does_not_establish":"Does not establish reliability on other machines or that the platform executed the run.","review_status":"passed","version":"suite-standard@1.0 · Q4_K_M","date":"2026-10-24T16:12:28.584Z"}],"updated_at":"2026-10-24T16:12:28.584Z","date":"2026-10-24","visibility":"public"},{"id":"3f14284b-55e8-4848-be61-2b379f278c6a","type":"run","title":"Qwen3-30B-A3B-Instruct-2507 on suite-standard@1.0","role":"Ran the suite","personal_contribution":"Ran suite-standard@1.0 on qwen3:30b-a3b-instruct-2507-q4_K_M, NVIDIA GeForce RTX 4090 · 24 GB.","outcome":"code-intervals 3/3 · tool-flight 3/3 · needle-8k 2/3 · 208.9 tok/s decode at speed@0","limitations":"Applies to this build, machine and suite version. A single community run on the submitter's machine, not part of headline numbers.","version":"suite-standard@1.0 · Q4_K_M","links":[],"disclosure":"not_stated","pinned":false,"run_id":"b683c8bc-8932-4903-a76b-83eda21b2cee","evidence":[{"type":"verified","label":"Verified","scope":"2 of 3 tasks regraded from the event log; code tasks not regraded","does_not_establish":"Does not establish reliability on other machines or that the platform executed the run.","review_status":"passed","version":"suite-standard@1.0 · Q4_K_M","date":"2026-10-21T01:24:36.301Z"}],"updated_at":"2026-10-21T01:24:36.301Z","date":"2026-10-21","visibility":"public"},{"id":"97cc1250-14e1-4658-badf-e42dd21cbf44","type":"video","title":"GLM-4.7-Flash 工具调用实测","role":null,"personal_contribution":null,"outcome":null,"limitations":null,"version":null,"links":[],"disclosure":"independent","pinned":false,"run_id":null,"evidence":[{"type":"source_linked","label":"Source linked","scope":"The link points to this creator's published work.","does_not_establish":"Does not establish the numbers in it.","review_status":"pending","version":null,"date":"2026-10-18T05:05:30.284Z"}],"updated_at":"2026-10-18T05:05:30.284Z","date":"2026-10-18","visibility":"public"},{"id":"726e3317-ec62-4998-83b9-bfb6f9047089","type":"run","title":"GLM-4.7-Flash on suite-standard@1.0","role":"Ran the suite","personal_contribution":"Ran suite-standard@1.0 on glm-4.7-flash:q4_K_M, Apple M4 Max · 128 GB.","outcome":"code-intervals 3/3 · tool-flight 1/3 · needle-8k 3/3 · 114.0 tok/s decode at speed@0","limitations":"Applies to this build, machine and suite version. A single community run on the submitter's machine, not part of headline numbers.","version":"suite-standard@1.0 · Q4_K_M","links":[],"disclosure":"not_stated","pinned":false,"run_id":"593b308c-e742-4a00-a081-adca7ab0ca46","evidence":[{"type":"verified","label":"Verified","scope":"2 of 3 tasks regraded from the event log; code tasks not regraded","does_not_establish":"Does not establish reliability on other machines or that the platform executed the run.","review_status":"passed","version":"suite-standard@1.0 · Q4_K_M","date":"2026-10-17T02:19:55.328Z"}],"updated_at":"2026-10-17T02:19:55.328Z","date":"2026-10-17","visibility":"public"},{"id":"05966523-fe3f-489e-80e6-d63c092d3a6d","type":"run","title":"gpt-oss-20b on suite-standard@1.0","role":"Ran the suite","personal_contribution":"Ran suite-standard@1.0 on gpt-oss:20b, NVIDIA GeForce RTX 4090 · 24 GB.","outcome":"code-intervals 2/3 · tool-flight 3/3 · needle-8k 3/3 · 201.0 tok/s decode at speed@0","limitations":"Applies to this build, machine and suite version. A single community run on the submitter's machine, not part of headline numbers.","version":"suite-standard@1.0 · MXFP4","links":[],"disclosure":"not_stated","pinned":false,"run_id":"92730a1d-81a0-4314-b64f-9c6233a566b0","evidence":[{"type":"verified","label":"Verified","scope":"2 of 3 tasks regraded from the event log; code tasks not regraded","does_not_establish":"Does not establish reliability on other machines or that the platform executed the run.","review_status":"passed","version":"suite-standard@1.0 · MXFP4","date":"2026-10-12T16:20:42.964Z"}],"updated_at":"2026-10-12T16:20:42.964Z","date":"2026-10-12","visibility":"public"},{"id":"ca6cec9b-02e6-4fa5-b85e-e8a428339f9b","type":"video","title":"4090 还是 M4 Max：本地跑 MoE 怎么选","role":null,"personal_contribution":null,"outcome":null,"limitations":null,"version":null,"links":[],"disclosure":"independent","pinned":false,"run_id":null,"evidence":[{"type":"source_linked","label":"Source linked","scope":"The link points to this creator's published work.","does_not_establish":"Does not establish the numbers in it.","review_status":"pending","version":null,"date":"2026-10-10T09:05:30.284Z"}],"updated_at":"2026-10-10T09:05:30.284Z","date":"2026-10-10","visibility":"public"},{"id":"a877fe37-3322-41c8-af25-378b8c3a7761","type":"run","title":"Llama-3.1-8B-Instruct on suite-standard@1.0","role":"Ran the suite","personal_contribution":"Ran suite-standard@1.0 on llama3.1:8b-instruct-q4_K_M, Apple M4 Max · 128 GB.","outcome":"code-intervals 2/3 · tool-flight 3/3 · needle-8k 2/3 · 71.9 tok/s decode at speed@0","limitations":"Applies to this build, machine and suite version. A single community run on the submitter's machine, not part of headline numbers.","version":"suite-standard@1.0 · Q4_K_M","links":[],"disclosure":"not_stated","pinned":false,"run_id":"975a9317-c91f-4abf-b45d-d62a727bdca3","evidence":[{"type":"verified","label":"Verified","scope":"2 of 3 tasks regraded from the event log; code tasks not regraded","does_not_establish":"Does not establish reliability on other machines or that the platform executed the run.","review_status":"passed","version":"suite-standard@1.0 · Q4_K_M","date":"2026-10-09T21:00:43.841Z"}],"updated_at":"2026-10-09T21:00:43.841Z","date":"2026-10-09","visibility":"public"},{"id":"9f6ca167-ede4-4466-ac10-f624f82ed703","type":"run","title":"gemma-3-27b-it on suite-standard@1.0","role":"Ran the suite","personal_contribution":"Queued suite-standard@1.0 for gemma3:27b-it-q4_K_M on BenchClub cloud, NVIDIA L40S · 48 GB, pinned environment.","outcome":"code-intervals 2/3 · tool-flight not supported · needle-8k 2/3 · 36.4 tok/s decode at speed@0","limitations":"Applies to this build and suite version on BenchClub's pinned cloud environment. Counts toward headline results.","version":"suite-standard@1.0 · Q4_K_M","links":[],"disclosure":"not_stated","pinned":false,"run_id":"f9b3ae65-4956-4a4d-9ef3-c325969f7da6","evidence":[{"type":"verified","label":"Verified","scope":"Run on BenchClub cloud; all 2 tasks graded by the platform","does_not_establish":"Does not establish results on other hardware.","review_status":"passed","version":"suite-standard@1.0 · Q4_K_M","date":"2026-10-07T08:29:53.691Z"}],"updated_at":"2026-10-07T08:29:53.691Z","date":"2026-10-07","visibility":"public"},{"id":"02fb8179-de0d-4cf8-94d5-0cc041bc6603","type":"run","title":"Mistral-7B-Instruct-v0.2 on suite-standard@1.0","role":"Ran the suite","personal_contribution":"Queued suite-standard@1.0 for mistral:7b-instruct-v0.2-q4_K_M on BenchClub cloud, NVIDIA L40S · 48 GB, pinned environment.","outcome":"code-intervals 2/3 · tool-flight not supported · needle-8k 2/3 · 136.8 tok/s decode at speed@0","limitations":"Applies to this build and suite version on BenchClub's pinned cloud environment. Counts toward headline results.","version":"suite-standard@1.0 · Q4_K_M","links":[],"disclosure":"not_stated","pinned":false,"run_id":"1dd9f6da-4efb-4b83-99a9-882429b15031","evidence":[{"type":"verified","label":"Verified","scope":"Run on BenchClub cloud; all 2 tasks graded by the platform","does_not_establish":"Does not establish results on other hardware.","review_status":"passed","version":"suite-standard@1.0 · Q4_K_M","date":"2026-10-07T08:11:35.809Z"}],"updated_at":"2026-10-07T08:11:35.809Z","date":"2026-10-07","visibility":"public"},{"id":"96d3f219-379e-4df1-b018-c4d3eca5061c","type":"run","title":"Qwen3-30B-A3B-Instruct-2507 on suite-standard@1.0","role":"Ran the suite","personal_contribution":"Ran suite-standard@1.0 on qwen3:30b-a3b-instruct-2507-q4_K_M, NVIDIA GeForce RTX 4090 · 24 GB.","outcome":"code-intervals 3/3 · tool-flight 2/3 · needle-8k 3/3 · 209.0 tok/s decode at speed@0","limitations":"Applies to this build, machine and suite version. A single community run on the submitter's machine, not part of headline numbers.","version":"suite-standard@1.0 · Q4_K_M","links":[],"disclosure":"not_stated","pinned":false,"run_id":"45a086b0-e2dd-4086-869b-374c8a9ab6d9","evidence":[{"type":"verified","label":"Verified","scope":"2 of 3 tasks regraded from the event log; code tasks not regraded","does_not_establish":"Does not establish reliability on other machines or that the platform executed the run.","review_status":"passed","version":"suite-standard@1.0 · Q4_K_M","date":"2026-10-05T19:28:30.742Z"}],"updated_at":"2026-10-05T19:28:30.742Z","date":"2026-10-05","visibility":"public"},{"id":"ff004967-8719-4b02-8c7b-1a99d70d0d6b","type":"run","title":"Qwen3-8B on suite-standard@1.0","role":"Ran the suite","personal_contribution":"Queued suite-standard@1.0 for qwen3:8b-q4_K_M on BenchClub cloud, NVIDIA L40S · 48 GB, pinned environment.","outcome":"code-intervals 2/3 · tool-flight 3/3 · needle-8k 3/3 · 122.4 tok/s decode at speed@0","limitations":"Applies to this build and suite version on BenchClub's pinned cloud environment. Counts toward headline results.","version":"suite-standard@1.0 · Q4_K_M","links":[],"disclosure":"not_stated","pinned":false,"run_id":"0044d963-410f-4e29-a076-fec51eebc573","evidence":[{"type":"verified","label":"Verified","scope":"Run on BenchClub cloud; all 3 tasks graded by the platform","does_not_establish":"Does not establish results on other hardware.","review_status":"passed","version":"suite-standard@1.0 · Q4_K_M","date":"2026-09-29T14:06:55.897Z"}],"updated_at":"2026-09-29T14:06:55.897Z","date":"2026-09-29","visibility":"public"}],"collaboration":{"formats":["Sponsored review","Live stream"],"availability":{"slots":2,"month":"December 2026","confirmed_at":"2026-11-02T11:05:30.284Z","unconfirmed":false}},"reputation":{"verified_works":16,"reproductions":0,"tier":"regular"},"updated_at":"2026-11-06T04:05:30.284Z"}