← variables.md

2026-08-26

71 change(s) in the model market

added to the catalog

• MiniMax: MiniMax M3 (free) (minimax/minimax-m3:free) added to the catalog: 1,048,576 ctx, $0/M in / $0/M out

removed from the catalog

• Google: Gemma 3n 4B (google/gemma-3n-e4b-it) removed from the catalog
• Qwen: Qwen Plus 0728 (thinking) (qwen/qwen-plus-2025-07-28:thinking) removed from the catalog

model price changes

• ~deepseek/deepseek-v4-flash-latest: prompt price $0.035/M → $0.03/M (−14.3%)
• ~moonshotai/kimi-latest: prompt price $2.8/M → $2.55/M (−8.9%)
• ~moonshotai/kimi-latest: completion price $14/M → $12.8/M (−8.9%)
• deepseek/deepseek-v4-flash-0731: prompt price $0.14/M → $0.06/M (−57.1%)
• deepseek/deepseek-v4-flash-0731: completion price $0.28/M → $0.12/M (−57.1%)
• deepseek/deepseek-v4-pro: prompt price $0.79/M → $0.87/M (+10.1%)
• deepseek/deepseek-v4-pro: completion price $1.58/M → $1.74/M (+10.1%)
• google/gemma-4-31b-it: prompt price $0.1/M → $0.09/M (−10.0%)
• meta-llama/llama-3.3-70b-instruct: prompt price $0.1/M → $0.71/M (+610.0%)
• meta-llama/llama-3.3-70b-instruct: completion price $0.32/M → $0.71/M (+121.9%)
• meta-llama/llama-4-scout: prompt price $0.1/M → $0.11/M (+10.0%)
• meta-llama/llama-4-scout: completion price $0.3/M → $0.34/M (+13.3%)
• qwen/qwen2.5-vl-72b-instruct: prompt price $0.8/M → $0.25/M (−68.8%)
• qwen/qwen2.5-vl-72b-instruct: completion price $1/M → $0.75/M (−25.0%)
• qwen/qwen3.5-397b-a17b: prompt price $0.5/M → $0.39/M (−22.0%)
• qwen/qwen3.5-397b-a17b: completion price $3.6/M → $2.34/M (−35.0%)
• qwen/qwen3.6-27b: prompt price $0.32/M → $0.6/M (+87.5%)
• qwen/qwen3.6-27b: completion price $3.2/M → $3.6/M (+12.5%)
• qwen/qwen3.8-27b: prompt price $0.4/M → $0.425/M (+6.3%)
• qwen/qwen3.8-27b: completion price $3/M → $2.55/M (−15.0%)

model context changes

• meta-llama/llama-guard-4-12b: context length 1,048,576 → 163,840

quantization changes

• GMICloud: deepseek/deepseek-v4-flash-0731 quantization fp4 → fp8
• Io Net: z-ai/glm-5.2 quantization fp8 → fp4

provider price changes

• Mancer 2: deepseek/deepseek-v4-flash-0731 prompt price $0.15/M → $0.155/M (+3.3%)
• OpenInference: deepseek/deepseek-v4-flash-0731 prompt price $0.035/M → $0.03/M (−14.3%)
• Relace: deepseek/deepseek-v4-flash-0731 prompt price $0.04/M → $0.06/M (+50.0%)
• Relace: deepseek/deepseek-v4-flash-0731 completion price $0.08/M → $0.12/M (+50.0%)
• StreamLake: deepseek/deepseek-v4-flash-0731 prompt price $0.088/M → $0.247/M (+180.7%)
• StreamLake: deepseek/deepseek-v4-flash-0731 completion price $0.064/M → $0.741/M (+1057.9%)
• StreamLake: deepseek/deepseek-v4-pro prompt price $0.79/M → $0.87/M (+10.1%)
• StreamLake: deepseek/deepseek-v4-pro completion price $1.58/M → $1.74/M (+10.1%)
• Decart: moonshotai/kimi-k2.6 prompt price $0.587/M → $0.568/M (−3.1%)
• Decart: moonshotai/kimi-k2.6 completion price $2.47/M → $3.33/M (+34.9%)
• Inceptron: moonshotai/kimi-k2.6 completion price $3.41/M → $3.39/M (−0.6%)
• AkashML: openai/gpt-oss-120b prompt price $0.037/M → $0.03/M (−18.9%)
• AkashML: openai/gpt-oss-120b completion price $0.49/M → $0.17/M (−65.3%)
• AkashML: qwen/qwen3.8-27b completion price $3/M → $2.55/M (−15.0%)
• DeepInfra: xiaomi/mimo-v2.5-pro prompt price $1/M → $0.39/M (−61.0%)
• DeepInfra: xiaomi/mimo-v2.5-pro completion price $3/M → $1.17/M (−61.0%)
• Inceptron: z-ai/glm-5.2 completion price $2.9/M → $2.4/M (−17.2%)
• Io Net: z-ai/glm-5.2 prompt price $1.26/M → $0.68/M (−46.0%)
• Io Net: z-ai/glm-5.2 completion price $4.4/M → $3.3/M (−25.0%)
• Novita: z-ai/glm-5.2 prompt price $0.7/M → $0.742/M (+6.0%)
• Novita: z-ai/glm-5.2 completion price $2.2/M → $2.33/M (+6.0%)
• StreamLake: z-ai/glm-5.2 prompt price $0.693/M → $0.735/M (+6.1%)
• StreamLake: z-ai/glm-5.2 completion price $2.18/M → $2.31/M (+6.1%)

provider context changes

• AkashML: deepseek/deepseek-v4-flash-0731 context length 131,072 → 1,048,576
• GMICloud: deepseek/deepseek-v4-flash-0731 context length 1,048,576 → 1,048,575

providers added

• Makora now serves deepseek/deepseek-v4-flash-0731 (unknown, 1,000,000 ctx, $0.09/M in / $0.195/M out)
• Makora now serves moonshotai/kimi-k3 (unknown, 1,048,576 ctx, $2.55/M in / $12.8/M out)
• Sail Research now serves moonshotai/kimi-k3 (fp4, 974,842 ctx, $2.6/M in / $13/M out)
• OpenAI now serves openai/gpt-5.1 (unknown, 400,000 ctx, $0.625/M in / $5/M out)
• OpenAI now serves openai/gpt-5.1 (unknown, 400,000 ctx, $1.25/M in / $10/M out)
• OpenAI now serves openai/gpt-5.1 (unknown, 400,000 ctx, $2.5/M in / $20/M out)
• OpenAI now serves openai/gpt-5.4 (unknown, 1,050,000 ctx, $5/M in / $30/M out)
• OpenAI now serves openai/gpt-5.4-mini (unknown, 400,000 ctx, $1.5/M in / $9/M out)
• Darkbloom now serves openai/gpt-oss-20b (fp8, 131,072 ctx, $0.02/M in / $0.1/M out)
• Upstage now serves upstage/solar-pro4 (unknown, 524,288 ctx, $0.03/M in / $0.12/M out)
• Amazon Bedrock now serves x-ai/grok-4.6 (unknown, 500,000 ctx, $2.2/M in / $6.6/M out)
• Makora now serves z-ai/glm-5.2 (fp4, 980,000 ctx, $0.72/M in / $2.38/M out)
• Makora now serves z-ai/glm-5.2 (fp8, 1,048,576 ctx, $0.94/M in / $2.8/M out)

providers removed

• OpenInference no longer serves google/gemma-4-31b-it (was bf16)
• Together no longer serves meta-llama/llama-guard-4-12b (was unknown)
• OpenAI no longer serves openai/gpt-5.1 (was unknown)
• OpenAI no longer serves openai/gpt-5.1 (was unknown)
• OpenAI no longer serves openai/gpt-5.1 (was unknown)
• OpenAI no longer serves openai/gpt-5.4 (was unknown)
• OpenAI no longer serves openai/gpt-5.4-mini (was unknown)

openrouter catalog snapshots 2026-08-25 → 2026-08-26 · schema 1 · generated 2026-08-26T05:50:35.400Z