← variables.md

2026-10-16

158 change(s) in the model market

added to the catalog

• Meta: Muse Spark Open (meta/muse-spark-open) added to the catalog: 1,048,576 ctx, $1.25/M in / $4.25/M out
• Moondream: Moondream 3 Preview (moondream/moondream3-preview) added to the catalog: 32,000 ctx, $0.3/M in / $2.5/M out
• Moondream: Moondream 3.1 (moondream/moondream3.1-9b-a2b) added to the catalog: 32,000 ctx, $0.3/M in / $2.5/M out
• NVIDIA: Nemotron 3.5 Super VL (nvidia/nemotron-3.5-super-vl) added to the catalog: 1,048,576 ctx, $0.15/M in / $0.5/M out
• NVIDIA: Nemotron 3.5 Super VL (free) (nvidia/nemotron-3.5-super-vl:free) added to the catalog: 1,000,000 ctx, $0/M in / $0/M out

removed from the catalog

• Anthropic: Claude Sonnet 4 (anthropic/claude-sonnet-4) removed from the catalog
• Pareto 26.10 Preview (unbiased/pareto-26.10-preview) removed from the catalog

model price changes

• ~deepseek/deepseek-flash-latest: prompt price $0.024/M → $0.036/M (+51.5%)
• ~deepseek/deepseek-flash-latest: completion price $0.72/M → $0.684/M (−5.0%)
• ~deepseek/deepseek-pro-latest: prompt price $0.225/M → $0.25/M (+11.1%)
• ~deepseek/deepseek-pro-latest: completion price $8/M → $5/M (−37.5%)
• ~deepseek/deepseek-v4-flash-latest: prompt price $0.014/M → $0.013/M (−9.0%)
• ~deepseek/deepseek-v4-flash-latest: completion price $0.576/M → $1.28/M (+122.2%)
• ~moonshotai/kimi-latest: prompt price $0.3/M → $0.294/M (−2.0%)
• ~moonshotai/kimi-latest: completion price $14/M → $24/M (+71.4%)
• ~z-ai/glm-flash-latest: prompt price $0.036/M → $0.021/M (−40.6%)
• ~z-ai/glm-flash-latest: completion price $0.9/M → $0.723/M (−19.6%)
• ~z-ai/glm-latest: prompt price $0.024/M → $0.061/M (+153.3%)
• deepseek/deepseek-chat: prompt price $0.32/M → $0.257/M (−19.6%)
• deepseek/deepseek-chat: completion price $0.89/M → $1.03/M (+15.6%)
• deepseek/deepseek-v4-flash-0731: prompt price $0.016/M → $0.013/M (−18.1%)
• deepseek/deepseek-v4-flash-0731: completion price $0.64/M → $1.28/M (+100.0%)
• deepseek/deepseek-v4.1-flash: prompt price $0.024/M → $0.036/M (+51.5%)
• deepseek/deepseek-v4.1-flash: completion price $0.72/M → $0.684/M (−5.0%)
• google/gemini-3.8-flash: prompt price $0.75/M → $0.825/M (+10.0%)
• google/gemini-3.8-flash: completion price $3.75/M → $4.13/M (+10.0%)
• meta-llama/llama-3.3-70b-instruct: prompt price $0.59/M → $0.1/M (−83.1%)
• meta-llama/llama-3.3-70b-instruct: completion price $0.79/M → $0.32/M (−59.5%)
• mistralai/mistral-small-3.2-24b-instruct: prompt price $0.075/M → $0.094/M (+25.0%)
• mistralai/mistral-small-3.2-24b-instruct: completion price $0.2/M → $0.25/M (+25.0%)
• moonshotai/kimi-k3: prompt price $0.45/M → $0.304/M (−32.4%)
• moonshotai/kimi-k3: completion price $15/M → $11.9/M (−20.4%)
• qwen/qwen3-next-80b-a3b-instruct: prompt price $0.1/M → $0.09/M (−10.0%)
• qwen/qwen3.6-27b: prompt price $0.32/M → $0.25/M (−21.9%)
• qwen/qwen3.6-27b: completion price $3.2/M → $2/M (−37.5%)
• z-ai/glm-5.2: prompt price $0.02/M → $0.037/M (+85.0%)
• z-ai/glm-5.2: completion price $7/M → $9/M (+28.6%)
• z-ai/glm-5.3: prompt price $0.03/M → $0.079/M (+163.3%)
• z-ai/glm-5.3-flash: prompt price $0.15/M → $0.021/M (−85.7%)
• z-ai/glm-5.3-flash: completion price $0.5/M → $0.723/M (+44.6%)
• z-ai/glm-5.3-flash:batch: prompt price $0.06/M → $0.05/M (−16.7%)
• z-ai/glm-5.3-flash:batch: completion price $0.2/M → $0.18/M (−10.0%)

quantization changes

• Wafer: deepseek/deepseek-v4-flash quantization unknown → mxfp4
• BaseTen: deepseek/deepseek-v4-flash-0731 quantization fp8 → fp32
• Wafer: deepseek/deepseek-v4-flash-0731 quantization unknown → nvfp4
• Wafer: deepseek/deepseek-v4-pro-0813 quantization unknown → mxfp4
• BaseTen: deepseek/deepseek-v4.1-flash quantization fp8 → unknown
• BaseTen: deepseek/deepseek-v4.1-flash quantization fp32 → unknown
• BaseTen: deepseek/deepseek-v4.1-flash quantization fp32 → unknown
• Wafer: deepseek/deepseek-v4.1-flash quantization unknown → mxfp4
• Parasail: google/gemma-3-4b-it quantization fp8 → bf16
• Wafer: moonshotai/kimi-k3 quantization unknown → mxfp4
• Wafer: moonshotai/kimi-k3 quantization unknown → mxfp4
• Wafer: qwen/qwen3.8-27b quantization unknown → fp8
• Wafer: z-ai/glm-5.2 quantization unknown → fp4
• Wafer: z-ai/glm-5.2 quantization unknown → fp4
• Wafer: z-ai/glm-5.3 quantization unknown → fp4
• Wafer: z-ai/glm-5.3 quantization unknown → fp4
• Wafer: z-ai/glm-5.3-flash quantization unknown → nvfp4

provider price changes

• Wafer: deepseek/deepseek-v4-flash prompt price $0.42/M → $0.58/M (+38.1%)
• Inceptron: deepseek/deepseek-v4-flash-0731 prompt price $0.13/M → $0.13/M (+0.3%)
• Inceptron: deepseek/deepseek-v4-flash-0731 completion price $0.6/M → $0.599/M (−0.1%)
• OpenInference: deepseek/deepseek-v4-flash-0731 prompt price $0.014/M → $0.012/M (−18.1%)
• OpenInference: deepseek/deepseek-v4-flash-0731 completion price $0.576/M → $1.15/M (+100.0%)
• Relace: deepseek/deepseek-v4-flash-0731 prompt price $0.016/M → $0.013/M (−18.1%)
• Relace: deepseek/deepseek-v4-flash-0731 completion price $0.64/M → $1.28/M (+100.0%)
• Relace: deepseek/deepseek-v4-pro-0813 prompt price $0.225/M → $0.6/M (+166.7%)
• Relace: deepseek/deepseek-v4-pro-0813 completion price $8/M → $3.8/M (−52.5%)
• AtlasCloud: deepseek/deepseek-v4.1-flash prompt price $0.192/M → $0.3/M (+56.2%)
• AtlasCloud: deepseek/deepseek-v4.1-flash completion price $0.768/M → $1.2/M (+56.2%)
• InferenceNet: deepseek/deepseek-v4.1-flash completion price $0.66/M → $0.76/M (+15.2%)
• Morph: deepseek/deepseek-v4.1-flash prompt price $0.033/M → $0.04/M (+21.2%)
• Morph: deepseek/deepseek-v4.1-flash completion price $0.6/M → $0.7/M (+16.7%)
• OpenInference: deepseek/deepseek-v4.1-flash prompt price $0.024/M → $0.036/M (+51.5%)
• OpenInference: deepseek/deepseek-v4.1-flash completion price $0.72/M → $0.684/M (−5.0%)
• Relace: deepseek/deepseek-v4.1-flash prompt price $0.026/M → $0.2/M (+657.6%)
• Relace: deepseek/deepseek-v4.1-flash completion price $0.8/M → $0.6/M (−25.0%)
• StreamLake: deepseek/deepseek-v4.1-flash prompt price $0.147/M → $0.15/M (+1.8%)
• StreamLake: deepseek/deepseek-v4.1-flash completion price $0.588/M → $0.599/M (+1.8%)
• Io Net: google/gemma-4-26b-a4b-it prompt price $0.11/M → $0.09/M (−18.2%)
• Io Net: google/gemma-4-26b-a4b-it completion price $0.32/M → $0.31/M (−3.1%)
• DeepInfra: google/gemma-4-31b-it prompt price $0.09/M → $0.135/M (+50.0%)
• DeepInfra: google/gemma-4-31b-it completion price $0.34/M → $0.51/M (+50.0%)
• DeepInfra: inclusionai/ling-3.1-flash prompt price $0.3/M → $0.21/M (−30.0%)
• DeepInfra: inclusionai/ling-3.1-flash completion price $0.9/M → $0.63/M (−30.0%)
• Inceptron: moonshotai/kimi-k2.6 completion price $2.45/M → $3.2/M (+30.6%)
• Decart: moonshotai/kimi-k3 prompt price $2.13/M → $1.65/M (−22.5%)
• Decart: moonshotai/kimi-k3 completion price $10.7/M → $8.25/M (−22.5%)
• InferenceNet: moonshotai/kimi-k3 prompt price $0.45/M → $0.304/M (−32.4%)
• InferenceNet: moonshotai/kimi-k3 completion price $15/M → $11.9/M (−20.4%)
• Morph: moonshotai/kimi-k3 prompt price $0.8/M → $1/M (+25.0%)
• Relace: moonshotai/kimi-k3 prompt price $0.32/M → $0.35/M (+9.4%)
• Wafer: moonshotai/kimi-k3 prompt price $0.31/M → $0.294/M (−5.2%)
• Wafer: moonshotai/kimi-k3 prompt price $0.33/M → $0.36/M (+9.1%)
• AtlasCloud: qwen/qwen3.6-35b-a3b prompt price $0.186/M → $0.25/M (+34.4%)
• AtlasCloud: qwen/qwen3.6-35b-a3b completion price $1.11/M → $1.25/M (+12.2%)
• Wafer: qwen/qwen3.8-27b prompt price $0.04/M → $0.2/M (+400.0%)
• Wafer: qwen/qwen3.8-27b completion price $2.3/M → $6/M (+160.9%)
• Makora: tencent/hy4-preview prompt price $0.83/M → $0.801/M (−3.5%)
• Makora: tencent/hy4-preview completion price $2.5/M → $2.48/M (−0.8%)
• Inceptron: z-ai/glm-5.2 completion price $3.15/M → $2.52/M (−20.0%)
• Relace: z-ai/glm-5.2 prompt price $0.019/M → $0.035/M (+84.7%)
• Relace: z-ai/glm-5.2 completion price $4/M → $6/M (+50.0%)
• Wafer: z-ai/glm-5.2 prompt price $0.02/M → $0.037/M (+85.0%)
• Wafer: z-ai/glm-5.2 completion price $7/M → $9/M (+28.6%)
• Wafer: z-ai/glm-5.2 completion price $8/M → $10/M (+25.0%)
• Decart: z-ai/glm-5.3 prompt price $0.92/M → $0.84/M (−8.7%)
• Decart: z-ai/glm-5.3 completion price $2.93/M → $2.66/M (−9.2%)
• InferenceNet: z-ai/glm-5.3 completion price $4.8/M → $4.66/M (−3.0%)
• Morph: z-ai/glm-5.3 prompt price $0.092/M → $0.109/M (+18.5%)
• Relace: z-ai/glm-5.3 prompt price $0.024/M → $0.061/M (+153.3%)
• Sail Research: z-ai/glm-5.3 prompt price $0.15/M → $0.5/M (+233.3%)
• Sail Research: z-ai/glm-5.3 completion price $3.4/M → $1.5/M (−55.7%)
• Wafer: z-ai/glm-5.3 prompt price $0.03/M → $0.17/M (+466.7%)
• Decart: z-ai/glm-5.3-flash prompt price $0.093/M → $0.096/M (+3.2%)
• Decart: z-ai/glm-5.3-flash completion price $0.31/M → $0.32/M (+3.2%)
• Inceptron: z-ai/glm-5.3-flash prompt price $0.108/M → $0.1/M (−7.4%)
• Inceptron: z-ai/glm-5.3-flash completion price $0.55/M → $0.5/M (−9.0%)
• InferenceNet: z-ai/glm-5.3-flash prompt price $0.083/M → $0.12/M (+44.6%)
• InferenceNet: z-ai/glm-5.3-flash completion price $0.15/M → $0.4/M (+166.7%)
• Morph: z-ai/glm-5.3-flash prompt price $0.069/M → $0.027/M (−60.9%)
• NextBit: z-ai/glm-5.3-flash prompt price $0.19/M → $0.25/M (+31.6%)
• NextBit: z-ai/glm-5.3-flash completion price $0.6/M → $0.7/M (+16.7%)
• OpenInference: z-ai/glm-5.3-flash prompt price $0.036/M → $0.019/M (−46.5%)
• Relace: z-ai/glm-5.3-flash prompt price $0.04/M → $0.021/M (−46.5%)
• Relace: z-ai/glm-5.3-flash completion price $0.5/M → $0.723/M (+44.6%)

providers added

• DeepInfra now serves deepseek/deepseek-chat (fp4, 163,840 ctx, $0.256/M in / $0.712/M out)
• Relace now serves deepseek/deepseek-v4-pro (fp4, 1,048,576 ctx, $0.9/M in / $3/M out)
• DekaLLM now serves deepseek/deepseek-v4-pro-0813 (unknown, 1,048,576 ctx, $0.88/M in / $3.8/M out)
• DeepInfra now serves deepseek/deepseek-v4.1-flash (fp8, 1,048,576 ctx, $0.112/M in / $0.336/M out)
• DeepInfra now serves deepseek/deepseek-v4.1-flash (fp8, 1,048,576 ctx, $0.21/M in / $0.63/M out)
• Google now serves google/gemini-3.1-flash-image (unknown, 131,072 ctx, $0.55/M in / $3.3/M out)
• Google now serves google/gemini-3.7-flash (unknown, 1,048,576 ctx, $0.825/M in / $4.13/M out)
• Google now serves google/gemini-3.8-flash (unknown, 1,048,576 ctx, $0.825/M in / $4.13/M out)
• DeepInfra now serves google/gemma-3-12b-it (bf16, 131,072 ctx, $0.04/M in / $0.12/M out)
• DeepInfra now serves google/gemma-3-12b-it (bf16, 131,072 ctx, $0.075/M in / $0.225/M out)
• DeepInfra now serves google/gemma-4-31b-it (fp8, 262,144 ctx, $0.16/M in / $0.32/M out)
• DeepInfra now serves google/gemma-4-31b-it (fp8, 262,144 ctx, $0.3/M in / $0.6/M out)
• DeepInfra now serves meta-llama/llama-4-scout (fp8, 327,680 ctx, $0.08/M in / $0.24/M out)
• DeepInfra now serves meta-llama/llama-4-scout (fp8, 327,680 ctx, $0.15/M in / $0.45/M out)
• DeepInfra now serves meta-llama/llama-guard-4-12b (bf16, 163,840 ctx, $0.144/M in / $0.144/M out)
• DeepInfra now serves meta-llama/llama-guard-4-12b (bf16, 163,840 ctx, $0.27/M in / $0.27/M out)
• DeepInfra now serves microsoft/phi-4 (bf16, 16,384 ctx, $0.056/M in / $0.112/M out)
• DeepInfra now serves minimax/minimax-m3 (fp8, 524,288 ctx, $0.224/M in / $0.88/M out)
• Crusoe now serves nvidia/nemotron-3-super-120b-a12b (fp8, 262,144 ctx, $0.3/M in / $2.4/M out)
• DeepInfra now serves nvidia/nemotron-3-ultra-550b-a55b (fp4, 262,144 ctx, $0.4/M in / $1.76/M out)
• Near AI now serves openai/gpt-oss-120b (fp4, 131,072 ctx, $0.03/M in / $0.15/M out)
• DeepInfra now serves qwen/qwen3-next-80b-a3b-instruct (fp8, 262,144 ctx, $0.072/M in / $0.88/M out)
• DeepInfra now serves qwen/qwen3-next-80b-a3b-instruct (fp8, 262,144 ctx, $0.135/M in / $1.65/M out)
• Featherless now serves stepfun/step-5-preview (fp4, 262,144 ctx, $1/M in / $2.7/M out)
• InferenceNet now serves stepfun/step-5-preview (fp4, 1,048,576 ctx, $1/M in / $2.7/M out)
• DeepInfra now serves z-ai/glm-4.6 (fp4, 202,752 ctx, $0.4/M in / $1.6/M out)
• DeepInfra now serves z-ai/glm-4.7 (fp4, 202,752 ctx, $0.32/M in / $1.4/M out)
• DeepInfra now serves z-ai/glm-5.2 (fp4, 1,048,576 ctx, $0.45/M in / $1.44/M out)
• DekaLLM now serves z-ai/glm-5.3 (unknown, 1,048,576 ctx, $0.18/M in / $2.95/M out)

providers removed

• Wafer no longer serves nvidia/nemotron-3.5-lightning (was unknown)
• DeepInfra no longer serves openai/gpt-oss-120b (was bf16)
• Tenstorrent no longer serves z-ai/glm-5.3 (was fp8)

openrouter catalog snapshots 2026-10-15 → 2026-10-16 · schema 1 · generated 2026-10-16T09:50:40.486Z