[BOT ISSUE] Missing Together AI models: Qwen/Qwen3.6-Plus, zai-org/GLM-5.1, MiniMaxAI/MiniMax-M2.7

[github-actions]

Summary

Three models are listed on Together AI's official serverless models page with pricing but are missing from the local catalog as Together models. All three already exist in the catalog under Fireworks with consistent specs, confirming the underlying models are real and active.

Gap Details

Missing Model ID	Together Pricing (in/out $/MTok)	Context Length	Fireworks Equivalent (already in catalog)
Qwen/Qwen3.6-Plus	$0.50 / $3.00	1,000,000	accounts/fireworks/models/qwen3p6-plus
zai-org/GLM-5.1	$1.40 / $4.40	202,752	accounts/fireworks/models/glm-5p1
MiniMaxAI/MiniMax-M2.7	$0.30 / $1.20 (cached: $0.06)	202,752	accounts/fireworks/models/minimax-m2p7

Verified Fields

Qwen/Qwen3.6-Plus

• Format: openai
• Flavor: chat
• Pricing: $0.50 input, $3.00 output per MTok (Together serverless page)
• Context: 1,000,000 tokens (Together serverless page)
• Function calling: Not listed as supported on Together
• displayName: "Qwen3.6 Plus" (matching Fireworks entry)

zai-org/GLM-5.1

• Format: openai
• Flavor: chat
• Pricing: $1.40 input, $4.40 output per MTok (Together serverless page)
• Context: 202,752 tokens (Together serverless page)
• Function calling: Yes (Together serverless page)
• Structured outputs: Yes (Together serverless page)
• reasoning: true (from Fireworks entry glm-5p1)
• max_output_tokens: 128,000 (from Fireworks entry)
• displayName: "GLM 5.1" (matching Fireworks entry)

MiniMaxAI/MiniMax-M2.7

• Format: openai
• Flavor: chat
• Pricing: $0.30 input, $1.20 output, $0.06 cached input per MTok (Together serverless page)
• Context: 202,752 tokens (Together serverless page)
• Function calling: Yes (Together serverless page)
• Structured outputs: Yes (Together serverless page)
• displayName: "MiniMax M2.7" (matching Fireworks entry)

Fields NOT verified

• max_output_tokens for Qwen3.6-Plus and MiniMax-M2.7 — not published on Together's serverless page
• max_input_tokens for Qwen3.6-Plus — Together shows 1M context but no explicit max_input_tokens field; downstream fix job should verify via API
• multimodal — not stated on Together's serverless page for any of these models
• reasoning for Qwen3.6-Plus and MiniMax-M2.7 — not stated on Together's serverless page

Official Sources

1. Together AI serverless models page: https://docs.together.ai/docs/serverless-models — confirms all three model IDs, pricing, and context lengths as available on Together's serverless inference
2. Fireworks catalog entries (already in model_list.json) independently confirm each underlying model exists:
   • accounts/fireworks/models/qwen3p6-plus (line ~3776)
   • accounts/fireworks/models/glm-5p1 (line ~9617)
   • accounts/fireworks/models/minimax-m2p7 (line ~10155)

Local Files Inspected

• packages/proxy/schema/model_list.json — grep for Qwen/Qwen3.6, zai-org/GLM-5.1, MiniMaxAI/MiniMax-M2.7 returns no matches
• Existing Together models follow the OrgName/ModelName convention (e.g., Qwen/Qwen3.5-397B-A17B at line ~4656)

Verification Notes

• Source 1 (Together serverless page) verifies: model IDs, pricing, context lengths, availability on Together
• Source 2 (Fireworks catalog) verifies: underlying models are real, active, and already tracked — pricing and specs are consistent across providers
• All three model IDs follow the exact OrgName/ModelName convention used by existing Together models in the catalog

{
  "kind": "missing_model",
  "provider": "together",
  "models": ["Qwen/Qwen3.6-Plus", "zai-org/GLM-5.1", "MiniMaxAI/MiniMax-M2.7"],
  "status": "active",
  "model_specs": {
    "Qwen/Qwen3.6-Plus": {
      "format": "openai",
      "flavor": "chat",
      "input_cost_per_mil_tokens": 0.5,
      "output_cost_per_mil_tokens": 3.0,
      "displayName": "Qwen3.6 Plus",
      "max_input_tokens": 1000000,
      "available_providers": ["together"]
    },
    "zai-org/GLM-5.1": {
      "format": "openai",
      "flavor": "chat",
      "input_cost_per_mil_tokens": 1.4,
      "output_cost_per_mil_tokens": 4.4,
      "displayName": "GLM 5.1",
      "reasoning": true,
      "max_input_tokens": 202752,
      "max_output_tokens": 128000,
      "available_providers": ["together"]
    },
    "MiniMaxAI/MiniMax-M2.7": {
      "format": "openai",
      "flavor": "chat",
      "input_cost_per_mil_tokens": 0.3,
      "output_cost_per_mil_tokens": 1.2,
      "input_cache_read_cost_per_mil_tokens": 0.06,
      "displayName": "MiniMax M2.7",
      "max_input_tokens": 202752,
      "available_providers": ["together"]
    }
  },
  "source_urls": [
    "https://docs.together.ai/docs/serverless-models",
    "https://fireworks.ai/pricing"
  ]
}

[github-actions]

Prepared catalog updates for Qwen/Qwen3.6-Plus, zai-org/GLM-5.1, MiniMaxAI/MiniMax-M2.7. Added models: Qwen/Qwen3.6-Plus, zai-org/GLM-5.1, MiniMaxAI/MiniMax-M2.7.

Verification sources: https://docs.together.ai/docs/serverless-models, https://fireworks.ai/pricing
Verified metadata applied:

• Qwen/Qwen3.6-Plus: format=openai, flavor=chat, displayName=Qwen3.6 Plus, max_input_tokens=1000000, max_output_tokens=not provided, pricing(input/output per 1M)=0.5/3, providers=together
• zai-org/GLM-5.1: format=openai, flavor=chat, displayName=GLM 5.1, max_input_tokens=202752, max_output_tokens=128000, pricing(input/output per 1M)=1.4/4.4, providers=together, reasoning=true
• MiniMaxAI/MiniMax-M2.7: format=openai, flavor=chat, displayName=MiniMax M2.7, max_input_tokens=202752, max_output_tokens=not provided, pricing(input/output per 1M)=0.3/1.2, input_cache_read_cost_per_mil_tokens=0.06, providers=together

Opened #521 to apply the catalog update.