2026-09-04
56 change(s) in the model market
added to the catalog
- Ling 3.0 Flash Fin (inclusionai/ling-3.0-flash-fin) added to the catalog: 262,144 ctx, $0.06/M in / $0.18/M out
- NVIDIA: Nemotron 3.5 Content Safety (nvidia/nemotron-3.5-content-safety) added to the catalog: 131,072 ctx, $0.2/M in / $0.2/M out
- SpaceXAI: Grok 4.3 (batch) (x-ai/grok-4.3:batch) added to the catalog: 1,000,000 ctx, $1/M in / $2/M out
model price changes
- ~moonshotai/kimi-latest: prompt price $2.55/M → $2.5/M (−2.0%)
- ~moonshotai/kimi-latest: completion price $12.8/M → $14/M (+9.8%)
- ~z-ai/glm-latest: prompt price $1.11/M → $1.15/M (+4.0%)
- ~z-ai/glm-latest: completion price $3.48/M → $3.5/M (+0.7%)
- deepseek/deepseek-chat: prompt price $0.257/M → $0.32/M (+24.3%)
- deepseek/deepseek-chat: completion price $1.03/M → $0.89/M (−13.5%)
- deepseek/deepseek-chat-v3.1: prompt price $0.25/M → $0.55/M (+120.0%)
- deepseek/deepseek-chat-v3.1: completion price $0.95/M → $1.65/M (+73.7%)
- nvidia/nemotron-3-ultra-550b-a55b: prompt price $0.6/M → $0.625/M (+4.2%)
- nvidia/nemotron-3-ultra-550b-a55b: completion price $2.4/M → $3.13/M (+30.2%)
- qwen/qwen2.5-vl-72b-instruct: prompt price $0.25/M → $0.8/M (+220.0%)
- qwen/qwen2.5-vl-72b-instruct: completion price $0.75/M → $1/M (+33.3%)
- qwen/qwen3.8-27b: prompt price $0.425/M → $0.42/M (−1.2%)
- qwen/qwen3.8-27b: completion price $2.55/M → $3/M (+17.6%)
- undi95/remm-slerp-l2-13b: prompt price $0.45/M → $0.35/M (−22.2%)
model context changes
- mistralai/mistral-medium-3-5:batch: context length 262,144 → 32,768
quantization changes
- GMICloud: deepseek/deepseek-v4-flash-0731 quantization fp4 → fp8
provider price changes
- Morph: deepseek/deepseek-v4-flash-0731 prompt price $0.099/M → $0.123/M (+25.0%)
- Morph: deepseek/deepseek-v4-flash-0731 completion price $0.278/M → $0.348/M (+25.0%)
- Novita: deepseek/deepseek-v4-flash-0731 prompt price $0.44/M → $0.409/M (−7.0%)
- Novita: deepseek/deepseek-v4-flash-0731 completion price $1.32/M → $1.23/M (−7.0%)
- Novita: deepseek/deepseek-v4-pro-0813 prompt price $1.32/M → $1.11/M (−16.0%)
- Novita: deepseek/deepseek-v4-pro-0813 completion price $3.96/M → $3.33/M (−16.0%)
- Decart: moonshotai/kimi-k2.6 prompt price $0.519/M → $0.568/M (+9.4%)
- Decart: moonshotai/kimi-k2.6 completion price $2.45/M → $3.33/M (+36.0%)
- Inceptron: moonshotai/kimi-k2.6 prompt price $0.53/M → $0.52/M (−1.9%)
- Morph: moonshotai/kimi-k3 prompt price $2.6/M → $2.5/M (−3.8%)
- Mancer 2: openai/gpt-oss-120b prompt price $0.06/M → $0.055/M (−8.3%)
- Together: qwen/qwen3.8-2.4t-a95b prompt price $2.5/M → $2/M (−20.0%)
- Together: qwen/qwen3.8-2.4t-a95b completion price $6.25/M → $6/M (−4.0%)
- AkashML: qwen/qwen3.8-27b prompt price $0.32/M → $0.25/M (−21.9%)
- AkashML: qwen/qwen3.8-27b completion price $2.5/M → $2.2/M (−12.0%)
- Parasail: qwen/qwen3.8-27b prompt price $0.25/M → $0.28/M (+12.0%)
- Reka: qwen/qwen3.8-27b prompt price $0.25/M → $0.24/M (−4.0%)
- Inceptron: z-ai/glm-5.2 prompt price $0.71/M → $1.25/M (+76.1%)
- Inceptron: z-ai/glm-5.2 completion price $2.35/M → $2.99/M (+27.2%)
- Decart: z-ai/glm-5.3 prompt price $1.11/M → $1.18/M (+6.3%)
- Decart: z-ai/glm-5.3 completion price $3.48/M → $3.7/M (+6.3%)
provider context changes
- GMICloud: deepseek/deepseek-v4-flash-0731 context length 1,048,576 → 1,048,575
providers added
- Sail Research now serves deepseek/deepseek-v4-flash-0731 (fp4, 1,048,576 ctx, $0.065/M in / $0.18/M out)
- SiliconFlow now serves deepseek/deepseek-v4-flash-vision-exp (fp8, 1,048,576 ctx, $0.44/M in / $1.32/M out)
- DeepInfra now serves inclusionai/ling-3.0-flash-fin (fp4, 262,144 ctx, $0.06/M in / $0.18/M out)
- Mistral now serves mistralai/mistral-large-2407 (unknown, 131,072 ctx, $2/M in / $6/M out)
- Mistral now serves mistralai/mistral-saba (unknown, 32,768 ctx, $0.2/M in / $0.6/M out)
- Mistral now serves mistralai/mixtral-8x22b-instruct (unknown, 65,536 ctx, $2/M in / $6/M out)
- Fireworks now serves moonshotai/kimi-k2.7-code (unknown, 262,144 ctx, $0.95/M in / $4/M out)
- DeepInfra now serves nvidia/nemotron-3.5-content-safety (bf16, 131,072 ctx, $0.2/M in / $0.2/M out)
- Darkbloom now serves qwen/qwen3.8-27b (fp4, 262,144 ctx, $0.22/M in / $2.42/M out)
providers removed
- Cloudflare no longer serves deepseek/deepseek-v4-flash (was unknown)
- Ambient no longer serves deepseek/deepseek-v4-flash-0731 (was fp4)
- Ambient no longer serves moonshotai/kimi-k2.7-code (was unknown)
- Nebius no longer serves qwen/qwen2.5-vl-72b-instruct (was fp8)
- Nebius no longer serves qwen/qwen3-32b (was fp8)
openrouter catalog snapshots 2026-09-03 → 2026-09-04 · schema 1 · generated 2026-09-04T09:40:13.761Z
$ curl variables.md — this site speaks markdown to your terminal.