[BOT ISSUE] Fireworks: fix GLM 5.2 cached input pricing from $0.26 to $0.14

[github-actions]

Provider

Fireworks

Model IDs

• accounts/fireworks/models/glm-5p2

Kind

stale_metadata

Current catalog values

{
  "input_cost_per_mil_tokens": 1.4,
  "output_cost_per_mil_tokens": 4.4,
  "input_cache_read_cost_per_mil_tokens": 0.26,
  "displayName": "GLM 5.2",
  "reasoning": true,
  "max_input_tokens": 1048576,
  "max_output_tokens": 131072,
  "available_providers": ["fireworks"]
}

Proposed update

{
  "input_cache_read_cost_per_mil_tokens": 0.14
}

Details

The Fireworks-specific entry accounts/fireworks/models/glm-5p2 has input_cache_read_cost_per_mil_tokens: 0.26, but Fireworks' official pricing page lists the cached input cost at $0.14 per million tokens. The input ($1.40) and output ($4.40) prices are correct.

The $0.26 value appears to come from Baseten's pricing for the same underlying model (zai-org/GLM-5.2 correctly has cache_read $0.26 per Baseten's pricing page). The Fireworks-specific entry should use Fireworks' own cache pricing.

For comparison, the older GLM 5.1 on Fireworks (accounts/fireworks/models/glm-5p1) correctly has cache_read $0.26, which matches Fireworks' official price for that model.

Sources

• Fireworks pricing page: https://docs.fireworks.ai/serverless/pricing — shows GLM 5.2 at $1.40 input / $0.14 cached / $4.40 output
• Verified twice via independent page fetches
• Baseten pricing (for comparison): https://www.baseten.co/pricing/ — shows GLM 5.2 cache at $0.26 (correctly used in zai-org/GLM-5.2)

Verification checklist

• Cross-source: Fireworks pricing page confirms $0.14 cached input (verified twice)
• Already-fixed: Catalog still shows $0.26, not $0.14
• ID format: accounts/fireworks/models/glm-5p2 already exists in model_list.json

[github-actions]

Prepared catalog updates for accounts/fireworks/models/glm-5p2. Updated models: accounts/fireworks/models/glm-5p2.

Verification sources: https://docs.fireworks.ai/serverless/pricing, https://www.baseten.co/pricing/
Verified metadata applied:

• accounts/fireworks/models/glm-5p2: format=openai, flavor=chat, displayName=GLM 5.2, max_input_tokens=1048576, max_output_tokens=131072, pricing(input/output per 1M)=1.4/4.4, input_cache_read_cost_per_mil_tokens=0.14, providers=fireworks, reasoning=true

Included in the daily batch PR: #904