diff options
Diffstat (limited to 'pi/agent/models.json.bak')
| -rw-r--r-- | pi/agent/models.json.bak | 743 |
1 files changed, 743 insertions, 0 deletions
diff --git a/pi/agent/models.json.bak b/pi/agent/models.json.bak new file mode 100644 index 0000000..f071691 --- /dev/null +++ b/pi/agent/models.json.bak @@ -0,0 +1,743 @@ +{ + "providers": { + "hyperstack": { + "baseUrl": "http://hyperstack1.wg1:11434/v1", + "apiKey": "$EMPTY", + "api": "openai-completions", + "compat": { + "supportsStore": false, + "supportsDeveloperRole": false, + "maxTokensField": "max_tokens" + }, + "models": [ + { + "id": "Qwen/Qwen3.6-27B-FP8", + "name": "Qwen3.6 27B FP8 [vm]", + "reasoning": true, + "input": [ + "text" + ], + "cost": { + "input": 0, + "output": 0, + "cacheRead": 0, + "cacheWrite": 0 + }, + "contextWindow": 262144, + "maxTokens": 8192, + "compat": { + "thinkingFormat": "qwen-chat-template", + "supportsStore": false, + "supportsDeveloperRole": false, + "maxTokensField": "max_tokens" + } + }, + { + "id": "cyankiwi/gemma-4-31B-it-AWQ-4bit", + "name": "Gemma 4 31B IT [vm]", + "reasoning": false, + "input": [ + "text" + ], + "cost": { + "input": 0, + "output": 0, + "cacheRead": 0, + "cacheWrite": 0 + }, + "contextWindow": 131072, + "maxTokens": 8192 + }, + { + "id": "cyankiwi/NVIDIA-Nemotron-3-Super-120B-A12B-AWQ-4bit", + "name": "Nemotron 3 Super 120B [vm]", + "reasoning": false, + "input": [ + "text" + ], + "cost": { + "input": 0, + "output": 0, + "cacheRead": 0, + "cacheWrite": 0 + }, + "contextWindow": 262144, + "maxTokens": 8192 + }, + { + "id": "Qwen/Qwen2.5-Coder-32B-Instruct-AWQ", + "name": "Qwen2.5 Coder 32B [vm]", + "reasoning": false, + "input": [ + "text" + ], + "cost": { + "input": 0, + "output": 0, + "cacheRead": 0, + "cacheWrite": 0 + }, + "contextWindow": 32768, + "maxTokens": 8192 + }, + { + "id": "QuantTrio/Qwen3-Coder-30B-A3B-Instruct-AWQ", + "name": "Qwen3 Coder 30B [vm]", + "reasoning": true, + "input": [ + "text" + ], + "cost": { + "input": 0, + "output": 0, + "cacheRead": 0, + "cacheWrite": 0 + }, + "contextWindow": 65536, + "maxTokens": 8192, + "compat": { + "thinkingFormat": "qwen-chat-template", + "supportsStore": false, + "supportsDeveloperRole": false, + "maxTokensField": "max_tokens" + } + }, + { + "id": "casperhansen/deepseek-r1-distill-qwen-32b-awq", + "name": "DeepSeek-R1-Distill 32B [vm]", + "reasoning": true, + "input": [ + "text" + ], + "cost": { + "input": 0, + "output": 0, + "cacheRead": 0, + "cacheWrite": 0 + }, + "contextWindow": 32768, + "maxTokens": 8192 + }, + { + "id": "Qwen/Qwen3-32B-AWQ", + "name": "Qwen3 32B [vm]", + "reasoning": true, + "input": [ + "text" + ], + "cost": { + "input": 0, + "output": 0, + "cacheRead": 0, + "cacheWrite": 0 + }, + "contextWindow": 32768, + "maxTokens": 8192, + "compat": { + "thinkingFormat": "qwen-chat-template", + "supportsStore": false, + "supportsDeveloperRole": false, + "maxTokensField": "max_tokens" + } + }, + { + "id": "cyankiwi/Devstral-Small-2507-AWQ-4bit", + "name": "Devstral Small 2507 [vm]", + "reasoning": false, + "input": [ + "text" + ], + "cost": { + "input": 0, + "output": 0, + "cacheRead": 0, + "cacheWrite": 0 + }, + "contextWindow": 32768, + "maxTokens": 8192 + } + ] + }, + "hyperstack1": { + "baseUrl": "http://hyperstack1.wg1:11434/v1", + "apiKey": "$EMPTY", + "api": "openai-completions", + "compat": { + "supportsStore": false, + "supportsDeveloperRole": false, + "maxTokensField": "max_tokens" + }, + "models": [ + { + "id": "cyankiwi/gemma-4-31B-it-AWQ-4bit", + "name": "Gemma 4 31B IT [vm1]", + "reasoning": false, + "input": [ + "text" + ], + "cost": { + "input": 0, + "output": 0, + "cacheRead": 0, + "cacheWrite": 0 + }, + "contextWindow": 131072, + "maxTokens": 8192 + }, + { + "id": "cyankiwi/NVIDIA-Nemotron-3-Super-120B-A12B-AWQ-4bit", + "name": "Nemotron 3 Super 120B 1M [vm1]", + "reasoning": false, + "input": [ + "text" + ], + "cost": { + "input": 0, + "output": 0, + "cacheRead": 0, + "cacheWrite": 0 + }, + "contextWindow": 1048576, + "maxTokens": 8192 + }, + { + "id": "Qwen/Qwen3.6-27B-FP8", + "name": "Qwen3.6 27B FP8 [vm1]", + "reasoning": true, + "input": [ + "text" + ], + "cost": { + "input": 0, + "output": 0, + "cacheRead": 0, + "cacheWrite": 0 + }, + "contextWindow": 262144, + "maxTokens": 8192, + "compat": { + "thinkingFormat": "qwen-chat-template", + "supportsStore": false, + "supportsDeveloperRole": false, + "maxTokensField": "max_tokens" + } + }, + { + "id": "Qwen/Qwen2.5-Coder-32B-Instruct-AWQ", + "name": "Qwen2.5 Coder 32B [vm1]", + "reasoning": false, + "input": [ + "text" + ], + "cost": { + "input": 0, + "output": 0, + "cacheRead": 0, + "cacheWrite": 0 + }, + "contextWindow": 32768, + "maxTokens": 8192 + }, + { + "id": "QuantTrio/Qwen3-Coder-30B-A3B-Instruct-AWQ", + "name": "Qwen3 Coder 30B [vm1]", + "reasoning": true, + "input": [ + "text" + ], + "cost": { + "input": 0, + "output": 0, + "cacheRead": 0, + "cacheWrite": 0 + }, + "contextWindow": 65536, + "maxTokens": 8192, + "compat": { + "thinkingFormat": "qwen-chat-template", + "supportsStore": false, + "supportsDeveloperRole": false, + "maxTokensField": "max_tokens" + } + }, + { + "id": "casperhansen/deepseek-r1-distill-qwen-32b-awq", + "name": "DeepSeek-R1-Distill 32B [vm1]", + "reasoning": true, + "input": [ + "text" + ], + "cost": { + "input": 0, + "output": 0, + "cacheRead": 0, + "cacheWrite": 0 + }, + "contextWindow": 32768, + "maxTokens": 8192 + }, + { + "id": "Qwen/Qwen3-32B-AWQ", + "name": "Qwen3 32B [vm1]", + "reasoning": true, + "input": [ + "text" + ], + "cost": { + "input": 0, + "output": 0, + "cacheRead": 0, + "cacheWrite": 0 + }, + "contextWindow": 32768, + "maxTokens": 8192, + "compat": { + "thinkingFormat": "qwen-chat-template", + "supportsStore": false, + "supportsDeveloperRole": false, + "maxTokensField": "max_tokens" + } + }, + { + "id": "cyankiwi/Devstral-Small-2507-AWQ-4bit", + "name": "Devstral Small 2507 [vm1]", + "reasoning": false, + "input": [ + "text" + ], + "cost": { + "input": 0, + "output": 0, + "cacheRead": 0, + "cacheWrite": 0 + }, + "contextWindow": 32768, + "maxTokens": 8192 + } + ] + }, + "hyperstack2": { + "baseUrl": "http://hyperstack2.wg1:11434/v1", + "apiKey": "$EMPTY", + "api": "openai-completions", + "compat": { + "supportsStore": false, + "supportsDeveloperRole": false, + "maxTokensField": "max_tokens" + }, + "models": [ + { + "id": "Qwen/Qwen3.6-27B-FP8", + "name": "Qwen3.6 27B FP8 [vm2]", + "reasoning": true, + "input": [ + "text" + ], + "cost": { + "input": 0, + "output": 0, + "cacheRead": 0, + "cacheWrite": 0 + }, + "contextWindow": 262144, + "maxTokens": 8192, + "compat": { + "thinkingFormat": "qwen-chat-template", + "supportsStore": false, + "supportsDeveloperRole": false, + "maxTokensField": "max_tokens" + } + }, + { + "id": "cyankiwi/gemma-4-31B-it-AWQ-4bit", + "name": "Gemma 4 31B IT [vm2]", + "reasoning": false, + "input": [ + "text" + ], + "cost": { + "input": 0, + "output": 0, + "cacheRead": 0, + "cacheWrite": 0 + }, + "contextWindow": 131072, + "maxTokens": 8192 + }, + { + "id": "Qwen/Qwen3.6-27B-FP8", + "name": "Qwen3.6 27B FP8 [vm2]", + "reasoning": true, + "input": [ + "text" + ], + "cost": { + "input": 0, + "output": 0, + "cacheRead": 0, + "cacheWrite": 0 + }, + "contextWindow": 262144, + "maxTokens": 8192, + "compat": { + "thinkingFormat": "qwen-chat-template", + "supportsStore": false, + "supportsDeveloperRole": false, + "maxTokensField": "max_tokens" + } + }, + { + "id": "cyankiwi/NVIDIA-Nemotron-3-Super-120B-A12B-AWQ-4bit", + "name": "Nemotron 3 Super 120B [vm2]", + "reasoning": false, + "input": [ + "text" + ], + "cost": { + "input": 0, + "output": 0, + "cacheRead": 0, + "cacheWrite": 0 + }, + "contextWindow": 262144, + "maxTokens": 8192 + }, + { + "id": "Qwen/Qwen2.5-Coder-32B-Instruct-AWQ", + "name": "Qwen2.5 Coder 32B [vm2]", + "reasoning": false, + "input": [ + "text" + ], + "cost": { + "input": 0, + "output": 0, + "cacheRead": 0, + "cacheWrite": 0 + }, + "contextWindow": 32768, + "maxTokens": 8192 + }, + { + "id": "QuantTrio/Qwen3-Coder-30B-A3B-Instruct-AWQ", + "name": "Qwen3 Coder 30B [vm2]", + "reasoning": true, + "input": [ + "text" + ], + "cost": { + "input": 0, + "output": 0, + "cacheRead": 0, + "cacheWrite": 0 + }, + "contextWindow": 65536, + "maxTokens": 8192, + "compat": { + "thinkingFormat": "qwen-chat-template", + "supportsStore": false, + "supportsDeveloperRole": false, + "maxTokensField": "max_tokens" + } + }, + { + "id": "casperhansen/deepseek-r1-distill-qwen-32b-awq", + "name": "DeepSeek-R1-Distill 32B [vm2]", + "reasoning": true, + "input": [ + "text" + ], + "cost": { + "input": 0, + "output": 0, + "cacheRead": 0, + "cacheWrite": 0 + }, + "contextWindow": 32768, + "maxTokens": 8192 + }, + { + "id": "Qwen/Qwen3-32B-AWQ", + "name": "Qwen3 32B [vm2]", + "reasoning": true, + "input": [ + "text" + ], + "cost": { + "input": 0, + "output": 0, + "cacheRead": 0, + "cacheWrite": 0 + }, + "contextWindow": 32768, + "maxTokens": 8192, + "compat": { + "thinkingFormat": "qwen-chat-template", + "supportsStore": false, + "supportsDeveloperRole": false, + "maxTokensField": "max_tokens" + } + }, + { + "id": "cyankiwi/Devstral-Small-2507-AWQ-4bit", + "name": "Devstral Small 2507 [vm2]", + "reasoning": false, + "input": [ + "text" + ], + "cost": { + "input": 0, + "output": 0, + "cacheRead": 0, + "cacheWrite": 0 + }, + "contextWindow": 32768, + "maxTokens": 8192 + } + ] + }, + "ollama": { + "baseUrl": "http://127.0.0.1:11434/v1", + "apiKey": "$EMPTY", + "api": "openai-completions", + "compat": { + "supportsStore": false, + "supportsDeveloperRole": false, + "maxTokensField": "max_tokens" + }, + "models": [ + { + "id": "kimi-k2.6:cloud", + "name": "kimi k2.6 1042B [ollama]", + "reasoning": true, + "input": [ + "text", + "image" + ], + "cost": { + "input": 0, + "output": 0, + "cacheRead": 0, + "cacheWrite": 0 + }, + "contextWindow": 262144, + "maxTokens": 8192 + }, + { + "id": "qwen3.5:cloud", + "name": "qwen3.5 397B [ollama]", + "reasoning": true, + "input": [ + "text", + "image" + ], + "cost": { + "input": 0, + "output": 0, + "cacheRead": 0, + "cacheWrite": 0 + }, + "contextWindow": 262144, + "maxTokens": 32768 + }, + { + "id": "glm-5.1:cloud", + "name": "glm-5.1 756B [ollama]", + "reasoning": true, + "input": [ + "text" + ], + "cost": { + "input": 0, + "output": 0, + "cacheRead": 0, + "cacheWrite": 0 + }, + "contextWindow": 202752, + "maxTokens": 131072 + }, + { + "id": "minimax-m2.7:cloud", + "name": "minimax-m2.7 229B [ollama]", + "reasoning": true, + "input": [ + "text" + ], + "cost": { + "input": 0, + "output": 0, + "cacheRead": 0, + "cacheWrite": 0 + }, + "contextWindow": 196608, + "maxTokens": 128000 + }, + { + "id": "qwen3-coder-next:cloud", + "name": "qwen3-coder-next 80B [ollama]", + "reasoning": true, + "input": [ + "text" + ], + "cost": { + "input": 0, + "output": 0, + "cacheRead": 0, + "cacheWrite": 0 + }, + "contextWindow": 262144, + "maxTokens": 32768 + }, + { + "id": "qwen3-coder:480b-cloud", + "name": "qwen3-coder 480B [ollama]", + "reasoning": false, + "input": [ + "text" + ], + "cost": { + "input": 0, + "output": 0, + "cacheRead": 0, + "cacheWrite": 0 + }, + "contextWindow": 262144, + "maxTokens": 32768 + }, + { + "id": "gpt-oss:20b-cloud", + "name": "gpt-oss 20B [ollama]", + "reasoning": true, + "input": [ + "text" + ], + "cost": { + "input": 0, + "output": 0, + "cacheRead": 0, + "cacheWrite": 0 + }, + "contextWindow": 131072, + "maxTokens": 16384 + }, + { + "id": "gpt-oss:120b-cloud", + "name": "gpt-oss 120B [ollama]", + "reasoning": true, + "input": [ + "text" + ], + "cost": { + "input": 0, + "output": 0, + "cacheRead": 0, + "cacheWrite": 0 + }, + "contextWindow": 131072, + "maxTokens": 16384 + }, + { + "id": "deepseek-v3.1:671b-cloud", + "name": "deepseek-v3.1 671B [ollama]", + "reasoning": true, + "input": [ + "text" + ], + "cost": { + "input": 0, + "output": 0, + "cacheRead": 0, + "cacheWrite": 0 + }, + "contextWindow": 163840, + "maxTokens": 16384 + }, + { + "id": "glm-4.6:cloud", + "name": "glm-4.6 357B [ollama]", + "reasoning": true, + "input": [ + "text" + ], + "cost": { + "input": 0, + "output": 0, + "cacheRead": 0, + "cacheWrite": 0 + }, + "contextWindow": 202752, + "maxTokens": 131072 + }, + { + "id": "minimax-m2:cloud", + "name": "minimax-m2 230B [ollama]", + "reasoning": false, + "input": [ + "text" + ], + "cost": { + "input": 0, + "output": 0, + "cacheRead": 0, + "cacheWrite": 0 + }, + "contextWindow": 204800, + "maxTokens": 128000 + } + ] + }, + "openrouter": { + "models": [ + { + "id": "qwen/qwen3.6-27b", + "name": "Qwen3.6 27B (dense)", + "reasoning": true, + "input": [ + "text", + "image" + ], + "cost": { + "input": 0.3, + "output": 3.2, + "cacheRead": 0, + "cacheWrite": 0 + }, + "contextWindow": 262144, + "maxTokens": 8192, + "compat": { + "thinkingFormat": "openrouter", + "supportsDeveloperRole": false, + "maxTokensField": "max_tokens" + } + }, + { + "id": "qwen/qwen3.6-35b-a3b", + "name": "Qwen3.6 35B-A3B (MoE)", + "reasoning": true, + "input": [ + "text", + "image" + ], + "cost": { + "input": 0.15, + "output": 1, + "cacheRead": 0, + "cacheWrite": 0 + }, + "contextWindow": 262144, + "maxTokens": 8192, + "compat": { + "thinkingFormat": "openrouter", + "supportsDeveloperRole": false, + "maxTokensField": "max_tokens", + "openRouterRouting": { + "order": [ + "AkashML" + ] + } + } + } + ] + } + } +} |