whatmodel.sh / Yi-Lightning
Yi-Lightning
↗ share
+ contribute
Yi-Lightning is a closed-weights language model from 01.AI, released October 2024.
Catalogue identity notes
Arena entry 'yi-lightning' (text). LMArena's leaderboard table (arena-leaderboard space @6f2eeeb) labels 'yi-lightning' as 'Yi-Lightning' (01 AI, Proprietary), linking https://platform.lingyiwanwu.com/docs#%E6%A8%A1%E5%9E%8B%E4%B8%8E%E8%AE%A1%E8%B4%B9. 01.AI's platform docs still list model 'yi-lightning' (16K context) in the model table and API examples (fetched 2026-09-28).
Model information sources
$ how it compares
The LLMs ranked either side of Yi-Lightning, out of 828 with a score, then more from 01.AI.
| # | model | score | where | ||
|---|---|---|---|---|---|
| 306 |  Magistral Medium 1.0 (2506) | 56±1 | comparison only | ||
| 307 |  GPT-4 Turbo (2024-04-09) | 55±0 | api $15.00/M | ||
| 308 | Yi-Lightning | 55±1 | api price unknown | ||
| 309 |  Qwen2.5 72B Instruct | 55±1 | api $0.37/M +local | ||
| 310 | Mistral Large 2 (2411) | 55±1 | local | ||
| more from 01.AI | |||||
| 373 | Yi-Large Preview | 52±1 | comparison only | ||
| 396 | Yi-Large | 51±1 | comparison only | ||
| 414 | Yi 34B 200K (base) | 50±5 | local | ||
$ what it's good at
coding▁▁▁▃▆██▇▆▄▃▁ #228 of 458 * · 53% of its score · 2 tests
Beats Qwen3.5 4B, Qwen3-VL 32B Instruct, Solar Open 100B (102B-A12B) here, though they rank higher overall. Trails QwQ-32B-Preview, Ministral 3 3B Reasoning 2512, GPT-4 (0613) here, though they rank lower overall.
| test | result | vs the others in its run | counts |
|---|---|---|---|
| Aider polyglot (225 exercises, two attempts) | 12.9% ✎ | ▂·▂▃▅▂▃█▆▄▂▂ beats 5/49 | 28% |
| Aider code editing (133 Exercism Python exercises, two attempts) | 65.4% ✎ | ▁·▁▂▄▁▆▄█▆▂▂ beats 34/62 | 25% |
chat & writing▁▁▁▃▅▄▅█▆▆▇▃ #182 of 393 * · 46% of its score · 2 tests
Beats Seed-OSS 36B Instruct, Nemotron-Cascade-14B-Thinking, Granite 4.2 8B here, though they rank higher overall. Trails LFM2.5 1.2B Instruct, LFM2.5 1.2B Thinking, Gemma 3 12B here, though they rank lower overall.
| test | result | vs the others in its run | counts |
|---|---|---|---|
| Arena-Hard (v0.1, GPT-4-1106 judge) | 81.5% ✎ | ▂▁▁▅█▅▄▃▃▅▃▁ beats 54/60 | 32% |
| LMArena (text) | 1328.3 ✎ | ▁▁▁▂▃▃▃▆▆▇█▃ beats 167/379 | 15% |
$ where to run it
Through an API
api · 01.AI (Lingyiwanwu platform) · price unknown · 16k context · platform.lingyiwanwu.com ↗
$ how the score was worked out · Arena-Hard-Auto v0.1 leaderbo… 32%, Aider polyglot benchmark runs 28%, Aider code editing benchmark… 25% and 1 more
Each result implies a score on its own, after adjusting for how hard its test is and who ran it. The score is their weighted average, with a small pull toward the middle for thinly tested models. Each test's weight is its counts column above; here it's added up by who ran the tests. How it works
| source | results | share of the score |
|---|---|---|
 Arena-Hard-Auto v0.1 leaderboard (repository README, pre-v2) | 1 | 32% |
 Aider polyglot benchmark runs | 1 | 28% |
| Aider code editing benchmark runs (old edit leaderboard) | 1 | 25% |
 LMArena leaderboard dataset | 1 | 15% |
| prior · pull toward the middle | 1% | |
| weighted average | = 55.4 |
$ improve this page
Use a short GitHub form or edit the data through a pull request. Every number needs a source; here's which ones we can use.
data updated · sources and licences · json