Models
A snapshot comparison of frontier and open language models — intelligence against price, speed, latency, and context — with the open-vs-closed and US-vs-China lines drawn in.
snapshot 2026-07-20 · 179 models · figures from Artificial Analysis, provider docs, and Epoch (ECI). Leaderboards churn — treat this as a dated picture, not live data.
intelligence vs blended price ($/M)
originlicense63 models
US China EU Canada Korea Israel open closed
top 16 · output speed (tok/s)
Mercury 21072
SWE-1.71000
LFM2.5-VL-1.6B418
Granite 4.0 H Small404
LFM2.5-8B-A1B337
Gemini 3.1 Flash-Lite333
gpt-oss-120b312
Nova Micro296
Nemotron 3 Nano Omni 30B A3B293
o3-mini236
Gemini 2.5 Flash235
Gemini 2.5 Flash-Lite216
Nemotron 3 Ultra 550B A55B209
gpt-oss-20b205
Qwen3.7 Max203
GPT-5.6 Luna199
US China
Are Chinese models catching up?
Intelligence against release date, with a best-so-far frontier line for each origin. The gap between the two envelopes — and whether it's closing — is the question the scaling01 analysis digs into.
the frontier over time · US vs China intelligence, best-so-far
US frontier China frontier open closed
Which models write alike?
Benchmarks measure what a model knows; this measures how it writes. Each model answered the same eight prompts, and the cells are the symmetric KL divergence between their character-trigram distributions — a fingerprint of style, from the words alone. Same-lab families cluster; the telling cells are the close ones across labs (shared training data leaves a stylistic trace). Reproduced from Typebulb's heat map.
who writes alike · symmetric KL over char-trigrams · 22 modelslower = more alike
Fable 5
Haiku 4.5
Opus 4.5
Opus 4.6
Opus 4.7
Opus 4.8
Sonnet 5
DeepSeek V4
Gemini 3 Flash
Gemini 3.1 Pro
Gemini 3.5 Flash
Kimi K2.6
Kimi K3
GPT 5.4
GPT 5.4 Mini
GPT 5.5
GPT 5.6 Luna
GPT 5.6 Sol
GPT 5.6 Terra
GLM 5.2
Grok 4.3
Grok 4.5
alikedistant0.42–0.96 bits
Hover a cell for the pair. Closest across labs: Zhipu: GLM 5.2 ~ Google: Gemini 3.1 Pro at 0.418 bits.
Prose only — each model answered eight fixed prompts; cells are the symmetric KL divergence between their character-trigram distributions (lower = writes more alike). Same-lab families cluster; the interesting cells are the low-divergence ones across labs. Data and method from Typebulb's “You're relatively right!” .
The full table
originlicense179 rows
| Model | Provider | Origin | Weights | AAI ↓ | $/M in | $/M out | $/M blend | tok/s | TTFT | Context | Size | Released |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Fable 5 | Anthropic | US | closed | 60 | $10 | $50 | $20 | 68 | 116.24s | 1M | — | 2026-06 |
| GPT-5.6 Sol | OpenAI | US | closed | 59 | $5 | $30 | $11.25 | 63 | 142.61s | 1M | — | 2026-07 |
| Kimi K3 | Moonshot | China | open | 57 | $3 | $15 | $6 | 62 | 1.99s | 1M | 2.8T | 2026-07 |
| Qwen3.7 Max | Alibaba | China | closed | 56.6 | $2.5 | $7.5 | $3.75 | 203 | 2.42s | 1M | — | 2026-05 |
| Opus 4.8 | Anthropic | US | closed | 56 | $5 | $25 | $10 | 60 | 30.89s | 1M | — | 2026-05 |
| GPT-5.5 | OpenAI | US | closed | 55 | $5 | $30 | $11.25 | 81 | 74.87s | 1.05M | — | 2026-04 |
| GPT-5.6 Terra | OpenAI | US | closed | 55 | $2.5 | $15 | $5.625 | 153 | 136.65s | 1M | — | 2026-07 |
| MiniMax M3 | MiniMax | China | open | 55 | $0.3 | $1.2 | $0.525 | 96 | 1.65s | 1M | 428B | 2026-06 |
| Grok 4.5 | xAI | US | closed | 54 | $2 | $6 | $3 | 70 | 10.27s | 500K | — | 2026-07 |
| Kimi K2.6 | Moonshot | China | open | 54 | $0.6 | $2.5 | $1.075 | — | — | 262K | 1.0T | 2026-04 |
| Opus 4.7 | Anthropic | US | closed | 54 | $5 | $25 | $10 | 53 | 20.46s | 1M | — | 2026-04 |
| Sonnet 5 | Anthropic | US | closed | 53 | $3 | $15 | $6 | 84 | 157.99s | 1M | — | 2026-06 |
| GLM-5.2 | Zhipu | China | open | 51 | $0.93 | $3 | $1.448 | 168 | — | 1M | 753B | 2026-06 |
| GPT-5.6 Luna | OpenAI | US | closed | 51 | $1 | $6 | $2.25 | 199 | 78.97s | 1M | — | 2026-07 |
| Muse Spark 1.1 | Meta | US | closed | 51 | $1.25 | $4.25 | $2 | 124 | 1.27s | 1M | — | 2026-07 |
| Gemini 3.5 Flash | US | closed | 50 | $1.5 | $9 | $3.375 | 165 | 20.31s | 1M | — | 2026-05 | |
| Nemotron 3 Ultra 550B A55B | NVIDIA | US | open | 47.7 | $0.675 | $2.675 | $1.175 | 209 | 1.23s | 262K | 550B | 2026-06 |
| Sonnet 4.6 | Anthropic | US | closed | 47 | $3 | $15 | $6 | 56 | 84.69s | 1M | — | 2026-02 |
| Gemini 3.1 Pro | US | closed | 46 | $2 | $12 | $4.5 | 129 | 33.44s | 1M | — | 2026-02 | |
| DeepSeek V4 Pro | DeepSeek | China | open | 44 | $0.435 | $0.87 | $0.544 | 62 | 1.77s | 1M | 1.6T | 2026-04 |
| GPT-5.3 Codex | OpenAI | US | closed | 44 | — | — | — | 143 | 62.98s | 400K | — | — |
| Opus 4.6 | Anthropic | US | closed | 44 | — | — | — | 44 | 19.49s | 1M | — | 2026-02 |
| Muse Spark | Meta | US | closed | 43 | — | — | — | — | — | 262K | — | 2026-04 |
| Hunyuan Hy3 | Tencent | China | open | 41 | $0.14 | $0.56 | $0.245 | — | — | 256K | 295B | 2026-07 |
| Gemini 3 Pro | US | closed | 40 | — | — | — | — | — | 1M | — | — | |
| Grok Build 0.1 | xAI | US | closed | 40 | $1 | $2 | $1.25 | 79 | 0.54s | 256K | — | 2026-06 |
| Gemini 3 Flash | US | closed | 38 | — | — | — | 190 | 7.03s | 1M | — | — | |
| Grok 4.3 | xAI | US | closed | 38 | $1.25 | $2.5 | $1.563 | 124 | 28.02s | 1M | — | 2026-04 |
| Grok 4.20 | xAI | US | closed | 37 | $1.25 | $2.5 | $1.563 | 188 | 11.11s | 2M | — | 2026-04 |
| Nemotron 3 Super 120B A12B | NVIDIA | US | open | 36 | $0.25 | $0.775 | $0.381 | 192 | 1.37s | 1M | 120B | 2026-03 |
| Grok 4 | xAI | US | closed | 33 | $5.5 | $27.5 | $11 | — | — | 256K | — | 2025-07 |
| Haiku 4.5 | Anthropic | US | closed | 30 | $1 | $5 | $2 | 105 | 16.21s | 200K | — | 2025-10 |
| Mistral Medium 3.5 | Mistral | EU | open | 30 | $1.5 | $7.5 | $3 | 124 | 2.09s | 256K | 128B | 2026-04 |
| o3 | OpenAI | US | closed | 30 | $2 | $8 | $3.5 | 159 | 7.07s | 200K | — | — |
| Step 3.7 Flash | StepFun | China | open | 30 | $0.2 | $1.15 | $0.438 | — | — | 262K | 196B | 2026-05 |
| Gemma 4 31B | US | open | 29 | $0.03 | — | — | 35 | 1.16s | 256K | 31B | 2026-04 | |
| GPT-5.5 Instant | OpenAI | US | closed | 29 | — | — | — | — | — | 400K | — | 2026-06 |
| Gemini 2.5 Pro | US | closed | 26 | $1.25 | $10 | $3.438 | 143 | 21.49s | 1M | — | — | |
| Gemma 4 26B A4B | US | open | 26 | $0.03 | — | — | 47 | 1.07s | 256K | 26B | 2026-04 | |
| Gemini 3.1 Flash-Lite | US | closed | 25 | $0.25 | $1.5 | $0.563 | 333 | 5.83s | 1M | — | 2026-05 | |
| gpt-oss-120b | OpenAI | US | open | 24 | $0.06 | — | — | 312 | 0.85s | 131K | 120B | — |
| EXAONE 4.5 33B | LG AI Research | Korea | open | 23 | — | — | — | — | — | 262K | 34.4B | 2026-04 |
| Gemma 4 12B | US | open | 22 | — | — | — | 110 | 2.54s | 256K | 12B | 2026-04 | |
| Apriel-1.6-15B-Thinker | ServiceNow | US | open | 21 | $0 | $0 | $0 | — | — | 128K | 15B | 2025-11 |
| Mercury 2 | Inception | US | closed | 21 | $0.25 | $0.75 | $0.375 | 1072 | 4.24s | 128K | — | 2026-02 |
| Gemini 2.5 Flash | US | closed | 20 | $0.3 | $2.5 | $0.85 | 235 | 14.95s | 1M | — | — | |
| Mistral Small 4 | Mistral | EU | open | 20 | $0.15 | $0.6 | $0.262 | 170 | 0.72s | 256K | 119B | 2026-03 |
| North Mini Code | Cohere | Canada | open | 20 | $0 | $0 | $0 | 119 | 0.35s | 256K | 30B | 2026-06 |
| Devstral 2 | Mistral | EU | open | 19 | $0 | $0 | $0 | 76 | 1.37s | 256K | 125B | 2025-12 |
| Hermes 4 405B | Nous Research | US | open | 19 | $1 | $3 | $1.5 | 39 | — | 130K | 405B | 2025-08 |
| o3-mini | OpenAI | US | closed | 19 | — | — | — | 236 | 6.44s | 200K | — | — |
| Magistral Medium 1.2 | Mistral | EU | closed | 18 | — | — | — | 42 | 1.8s | 131K | — | — |
| Sarvam 105B | Sarvam AI | India | open | 18 | — | — | — | — | — | 128K | 106B | 2026-04 |
| Sonar Reasoning Pro | Perplexity | US | closed | 18 | $2 | $8 | $3.5 | — | — | 128K | — | 2025-01 |
| EXAONE 4.0 32B | LG AI Research | Korea | open | 17 | $0.6 | $1 | $0.7 | — | — | — | 32B | 2025-07 |
| Grok 4.1 Fast | xAI | US | closed | 17 | $0.2 | $0.5 | $0.275 | — | — | 2M | — | 2025-11 |
| HyperCLOVA X SEED Think 32B | Naver | Korea | open | 17 | — | — | — | — | — | 128K | 32B | 2025-12 |
| Grok 4 Fast | xAI | US | closed | 16 | $0.2 | $0.5 | $0.275 | — | — | 2M | — | 2025-09 |
| Mistral Large 3 | Mistral | EU | open | 16 | $0.5 | $1.5 | $0.75 | 51 | 1.12s | 260K | 675B | 2025-12 |
| gpt-oss-20b | OpenAI | US | open | 15 | $0.02 | — | — | 205 | 0.79s | 131K | 20B | — |
| Mistral Medium 3.1 | Mistral | EU | closed | 15 | $0.4 | $2 | $0.8 | 90 | 1.52s | 128K | — | 2025-08 |
| Nemotron 3 Nano Omni 30B A3B | NVIDIA | US | open | 15 | $0.07 | $0.3 | $0.128 | 293 | — | 256K | 30B | 2026-04 |
| Llama 4 Maverick | Meta | US | open | 14 | $0.35 | $0.85 | $0.475 | 113 | 0.97s | 1M | 402B | 2025-04 |
| Nova 2.0 Pro Preview | Amazon | US | closed | 14 | $1.25 | $10 | $3.438 | 114 | 1.1s | 256K | — | 2025-11 |
| Solar Pro 3 | Upstage | Korea | closed | 14 | — | — | — | — | — | — | — | 2026-04 |
| Nova Premier | Amazon | US | closed | 13 | $2.5 | $12.5 | $5 | 28 | 2.98s | 1M | — | 2025-04 |
| Gemma 4 E4B | US | open | 12 | — | — | — | 94 | 1.13s | 128K | 4B | 2026-04 | |
| Nova 2.0 Lite | Amazon | US | closed | 12 | $0.3 | $2.5 | $0.85 | 159 | 1.03s | 1M | — | 2025-10 |
| OLMo 3 32B Think | AI2 | US | open | 12 | $0.15 | $0.5 | $0.238 | — | — | 64K | 32B | 2025-11 |
| Sarvam 30B | Sarvam AI | India | open | 12 | — | — | — | — | — | 65K | 32B | 2026-04 |
| Gemini 2.5 Flash-Lite | US | closed | 11 | $0.1 | $0.4 | $0.175 | 216 | 0.37s | 1M | — | — | |
| Granite 4.0 H Small | IBM | US | open | 11 | $0.06 | $0.25 | $0.108 | 404 | — | 130K | 32B | 2025-10 |
| Llama 4 Scout | Meta | US | open | 10 | $0.175 | $0.625 | $0.288 | 87 | 0.8s | 10M | 109B | 2025-04 |
| Phi-4 Multimodal | Microsoft | US | open | 10 | — | — | — | 17 | — | 128K | — | — |
| Sonar | Perplexity | US | closed | 10 | $1 | $1 | $1 | — | — | 128K | — | 2025-01 |
| Gemma 4 E2B | US | open | 9 | — | — | — | — | — | 128K | 2B | 2026-04 | |
| Granite 4.1 30B | IBM | US | open | 9 | — | — | — | — | — | 512K | 30B | 2026-04 |
| Granite 4.1 3B | IBM | US | open | 9 | — | — | — | — | — | 512K | 3B | 2026-04 |
| Sonar Pro | Perplexity | US | closed | 9 | $3 | $15 | $6 | — | — | 200K | — | 2025-01 |
| Command A | Cohere | Canada | open | 8 | $2.5 | $10 | $4.375 | 63 | 1.68s | 256K | 111B | 2025-03 |
| LFM2.5-8B-A1B | Liquid AI | US | open | 8 | — | — | — | 337 | 2.92s | 33K | 8B | — |
| Nova Pro | Amazon | US | closed | 8 | $0.8 | $3.2 | $1.4 | — | — | 300K | — | 2024-12 |
| OLMo 3 7B Instruct | AI2 | US | open | 8 | $0.1 | $0.2 | $0.125 | — | — | 64K | 7B | 2025-11 |
| OLMo 3.1 32B Think | AI2 | US | open | 8 | $0 | $0 | $0 | — | — | 64K | 32B | 2026-01 |
| Sarvam-M | Sarvam AI | India | open | 8 | — | — | — | — | — | 33K | 24B | 2025-05 |
| Gemma 3 27B | US | open | 7 | — | — | — | — | — | 128K | 27B | — | |
| Granite 4.1 8B | IBM | US | open | 7 | — | — | — | — | — | 512K | 8B | 2026-04 |
| Nemotron 3 Nano 30B A3B | NVIDIA | US | open | 7 | $0.05 | $0.2 | $0.088 | 115 | 0.96s | 1M | 30B | 2025-12 |
| Nova Lite | Amazon | US | closed | 7 | $0.06 | $0.24 | $0.105 | 157 | 0.94s | 300K | — | 2024-12 |
| Jamba 1.7 Large | AI21 | Israel | open | 5 | $2 | $8 | $3.5 | 57 | 1.63s | 256K | 398B | 2025-07 |
| Nova Micro | Amazon | US | closed | 5 | $0.035 | $0.14 | $0.061 | 296 | 0.86s | 128K | — | 2024-12 |
| Phi-4 | Microsoft | US | open | 5 | $0.125 | $0.5 | $0.219 | — | — | 16K | 14B | 2024-12 |
| Reka Flash 3 | Reka | US | open | 4 | $0.2 | $0.8 | $0.35 | — | — | 128K | 21B | 2025-03 |
| Jamba 1.6 Mini | AI21 | Israel | open | 3 | — | — | — | 181 | 0.76s | 256K | — | — |
| Jamba Mini 1.7 | AI21 | Israel | open | 3 | $0.2 | $0.4 | $0.25 | — | — | 256K | 52B | 2025-07 |
| LFM2-2.6B | Liquid AI | US | open | 3 | — | — | — | — | — | 33K | 2.6B | 2025-09 |
| Gemma 3 270M | US | open | 2 | — | — | — | — | — | 32K | 0.27B | — | |
| LFM2.5-VL-1.6B | Liquid AI | US | open | 1 | — | — | — | 418 | 4.04s | 33K | 1.6B | — |
| Agents-A1 | InternScience | China | open | — | — | — | — | — | — | 262K | 35B | — |
| Apple On-Device Foundation | Apple | US | closed | — | — | — | — | — | — | — | 3B | 2025-06 |
| Audex | NVIDIA | US | open | — | — | — | — | — | — | 1M | 30B | — |
| Bonsai 27B | PrismML | Other | open | — | — | — | — | 163 | — | 262K | 27B | — |
| Codestral Mamba 7B | Mistral AI | EU | open | — | — | — | — | — | — | 256K | 7B | 2024-07 |
| Command R+ (08-2024) | Cohere | Canada | open | — | $2.5 | $10 | $4.375 | — | — | 128K | 104B | 2024-08 |
| Command R7B | Cohere | Canada | open | — | $0.0375 | $0.15 | $0.066 | — | — | 128K | 7B | 2024-12 |
| Cosmos 3 | NVIDIA | US | open | — | — | — | — | — | — | — | 64B | — |
| DBRX Instruct | Databricks | US | open | — | — | — | — | — | — | 32K | 132B | 2024-03 |
| DeepSeek V4 Flash | DeepSeek | China | open | — | $0.14 | $0.28 | $0.175 | — | — | 1M | 284B | 2026-04 |
| DeepSeek-R1-Distill-Llama-70B | DeepSeek | China | open | — | — | — | — | — | — | 128K | 70B | 2025-01 |
| DeepSeek-R1-Distill-Qwen-32B | DeepSeek | China | open | — | — | — | — | — | — | 128K | 32B | 2025-01 |
| ERNIE 4.5 300B-A47B | Baidu | China | open | — | — | — | — | — | — | 128K | 300B | 2025-06 |
| ERNIE 5.1 | Baidu | China | closed | — | $0.59 | $2.65 | $1.105 | — | — | 128K | — | 2026-05 |
| Falcon-180B | TII | UAE | open | — | — | — | — | — | — | 2K | 180B | 2023-09 |
| Falcon-H1-34B | TII | UAE | open | — | — | — | — | — | — | 256K | 34B | 2025-05 |
| Gemma 4 | US | open | — | — | — | — | — | — | 128K | 31B | — | |
| Hermes 4 70B | Nous Research | US | open | — | — | — | — | — | — | 131K | 70B | 2025-08 |
| iFlytek Spark X1 | iFlytek | China | closed | — | — | — | — | — | — | — | — | — |
| iLLaDA | Renmin University & ByteDance | China | closed | — | — | — | — | — | — | 8K | 7.62B | — |
| Inkling | Thinking Machines Lab | US | open | — | — | — | — | — | — | 1M | 975B | — |
| Intern-S2 | InternLM | China | open | — | — | — | — | — | — | 256K | 397B | — |
| KAT-Coder-V2.5 | Kwaipilot | China | closed | — | — | — | — | — | — | — | — | — |
| Laguna M.1 | Poolside | US | closed | — | — | — | — | — | — | 256K | 225.8B | — |
| Leanstral 1.5 | Mistral AI | EU | open | — | — | — | — | — | — | — | 119B | — |
| Llama 3.1 405B | Meta | US | open | — | — | — | — | — | — | 128K | 405B | 2024-07 |
| Llama 3.1 8B | Meta | US | open | — | — | — | — | — | — | 128K | 8B | 2024-07 |
| Llama 3.2 3B | Meta | US | open | — | — | — | — | — | — | 128K | 3B | 2024-09 |
| Llama 3.3 70B | Meta | US | open | — | — | — | — | — | — | 128K | 70B | 2024-12 |
| Llama 4 Behemoth | Meta | US | closed | — | — | — | — | — | — | — | 2.0T | — |
| Llama-3.1-Nemotron-Ultra-253B | NVIDIA | US | open | — | — | — | — | — | — | 128K | 253B | 2025-04 |
| Llama-3.3-Nemotron-Super-49B v1.5 | NVIDIA | US | open | — | — | — | — | — | — | 128K | 49B | 2025-07 |
| LongCat 2.0 | Meituan | China | open | — | — | — | — | — | — | 1M | 1.6T | — |
| LOTUS | academic | US | closed | — | — | — | — | — | — | — | 3B | — |
| Mach-Mind-4-Flash | Li Auto | China | closed | — | — | — | — | — | — | — | 35B | — |
| MAI-1-preview | Microsoft AI | US | closed | — | — | — | — | — | — | — | — | 2025-08 |
| MAI-Thinking-1 | Microsoft AI | US | closed | — | — | — | — | — | — | 256K | — | 2026-05 |
| MiMo-V2-Flash | Xiaomi | China | open | — | — | — | — | — | — | 256K | 309B | — |
| Mistral NeMo 12B | Mistral AI | EU | open | — | — | — | — | — | — | 128K | 12B | 2024-07 |
| Mistral Small 3.2 | Mistral AI | EU | open | — | — | — | — | — | — | 128K | 24B | 2025-06 |
| Mixtral 8x22B | Mistral AI | EU | open | — | — | — | — | — | — | 64K | 141B | 2024-04 |
| Mixtral 8x7B | Mistral AI | EU | open | — | — | — | — | — | — | 32K | 46.7B | 2023-12 |
| Monolith 1.0 | Basalt Labs | Other | open | — | — | — | — | — | — | 1M | 1.6T | — |
| Motif 2.6B | Motif Technologies | Korea | open | — | — | — | — | — | — | 16K | 2.6B | — |
| MusaCoder | Moore Threads | China | closed | — | — | — | — | — | — | — | 27B | — |
| Nemotron Nano 9B v2 | NVIDIA | US | open | — | — | — | — | — | — | 128K | 9B | 2025-08 |
| Nemotron NVFP4 | NVIDIA | US | closed | — | — | — | — | — | — | — | 550B | — |
| Nemotron TwoTower | NVIDIA | US | open | — | — | — | — | — | — | — | 30B | — |
| Nemotron-H 56B | NVIDIA | US | open | — | — | — | — | — | — | 8K | 56B | 2025-04 |
| Palmyra X5 | Writer | US | closed | — | — | — | — | — | — | 1M | — | 2025-04 |
| Phi-3.5-MoE | Microsoft | US | open | — | — | — | — | — | — | 128K | 42B | 2024-08 |
| Phi-4-reasoning-plus | Microsoft | US | open | — | — | — | — | — | — | 32K | 14B | 2025-04 |
| Qwen-Audio-3.0-TTS | Alibaba | China | closed | — | — | — | — | — | — | — | — | — |
| Qwen2.5-72B | Alibaba | China | open | — | — | — | — | — | — | 128K | 72B | 2024-09 |
| Qwen2.5-Coder-32B | Alibaba | China | open | — | — | — | — | — | — | 128K | 32B | 2024-11 |
| Qwen3-14B | Alibaba | China | open | — | — | — | — | — | — | 128K | 14B | 2025-04 |
| Qwen3-235B-A22B | Alibaba | China | open | — | — | — | — | — | — | 128K | 235B | 2025-04 |
| Qwen3-30B-A3B | Alibaba | China | open | — | — | — | — | — | — | 128K | 30B | 2025-04 |
| Qwen3-32B | Alibaba | China | open | — | — | — | — | — | — | 128K | 32B | 2025-04 |
| Qwen3-4B | Alibaba | China | open | — | — | — | — | — | — | 32K | 4B | 2025-04 |
| Qwen3-8B | Alibaba | China | open | — | — | — | — | — | — | 128K | 8B | 2025-04 |
| Qwen3-Coder-480B-A35B | Alibaba | China | open | — | — | — | — | — | — | 256K | 480B | 2025-07 |
| Qwen3-VL-235B-A22B | Alibaba | China | open | — | — | — | — | — | — | 256K | 235B | 2025-09 |
| Qwen3.8 Max | Alibaba | China | closed | — | — | — | — | — | — | — | 2.4T | 2026-07 |
| QwQ-32B | Alibaba | China | open | — | — | — | — | — | — | 128K | 32B | 2025-03 |
| Ring-Zero | Ant Group | China | closed | — | — | — | — | — | — | 64K | 1.0T | — |
| Sakana Fugu | Sakana AI | Other | closed | — | — | — | — | — | — | — | — | — |
| SmolLM3-3B | Hugging Face | US | open | — | — | — | — | — | — | 128K | 3B | 2025-07 |
| Snowflake Arctic | Snowflake | US | open | — | — | — | — | — | — | 4K | 480B | 2024-04 |
| Solar Pro 2 | Upstage | Korea | open | — | — | — | — | — | — | — | 31B | — |
| SOLAR-10.7B | Upstage | Korea | open | — | — | — | — | — | — | 4K | 10.7B | 2023-12 |
| Soofi S | KI Bundesverband | EU | open | — | — | — | — | — | — | 1M | 31.6B | — |
| SWE-1.7 | Cognition | US | closed | — | — | — | — | 1000 | — | — | — | 2026-07 |
| TabFM | US | open | — | — | — | — | — | — | — | — | — | |
| Tapered LM | academic | Canada | closed | — | — | — | — | — | — | — | 1.3B | — |
| Un-0 | Unconventional AI | US | open | — | — | — | — | — | — | — | 0.322B | — |
| VideoChat3 | Shanghai AI Lab | China | open | — | — | — | — | — | — | — | 4B | — |
| Yi-1.5-34B | 01.AI | China | open | — | — | — | — | — | — | 32K | 34B | 2024-05 |
| Yi-Lightning | 01.AI | China | closed | — | — | — | — | — | — | — | — | 2024-10 |
| Zamba2-7B | Zyphra | US | open | — | — | — | — | — | — | — | 7B | 2024-10 |
| ZUNA 1.1 | Zyphra | US | open | — | — | — | — | — | — | — | 0.38B | 2026-07 |