Skip to content

feat(pricing): add OpenCode Go model entries to catalog#7

Merged
bravo1goingdark merged 1 commit into
mainfrom
feat/opencode-pricing
May 5, 2026
Merged

feat(pricing): add OpenCode Go model entries to catalog#7
bravo1goingdark merged 1 commit into
mainfrom
feat/opencode-pricing

Conversation

@bravo1goingdark
Copy link
Copy Markdown
Owner

Summary

Adds 14 OpenCode Go model slugs to the bundled LiteLLM-derived pricing catalog. OpenCode's curated set isn't in upstream LiteLLM yet (their slugs lag releases by a few weeks), so we use sibling-family rates as documented approximations.

OpenCode slug Rate source $/M in / $/M out
`opencode/glm-5.1`, `opencode/glm-5` `baseten/zai-org/GLM-5` 0.95 / 3.15
`opencode/kimi-k2.6`, `kimi-k2.5` `bedrock/moonshotai.kimi-k2.5` 0.60 / 3.03
`opencode/minimax-m2.7`, `m2.5` `bedrock/minimax.minimax-m2.5` 0.36 / 1.44
`opencode/deepseek-v4-{pro,flash}` `azure_ai/deepseek-v3.2` 0.58 / 1.68
`opencode/qwen3.{5,6}-plus` `bedrock/qwen.qwen3-coder-next` 0.60 / 1.50
`opencode/mimo-v2-{pro,omni,2.5,2.5-pro}` `openrouter/xiaomi/mimo-v2-flash` 0.09 / 0.29

Keys are prefixed `opencode/` to group with the rest of the provider's catalog, mirroring how mistral/groq entries are stored. The catalog's existing prefix-strip during lookup resolves the bare slug.

Cost reported is upstream-equivalent value, not what the user actually pays — OpenCode Go is a flat-rate $10/mo plan. Useful for tracking $60/mo cap consumption and comparing against direct-provider pricing. Obol's docs will call this out as approximate.

Test plan

  • `cargo test -p keplor-pricing` — 12 passed
  • `cargo build --release --bin keplor` clean
  • After deploy: fire glm-5.1 request via Obol, query `/v1/events`, confirm `cost_nanodollars` is non-zero

Adds 14 OpenCode Go model slugs to the bundled LiteLLM-derived
pricing catalog. OpenCode's curated set isn't in upstream LiteLLM
yet (their slugs lag releases by a few weeks), so we use
sibling-family rates as documented approximations:

  glm-5.1, glm-5         from baseten/zai-org/GLM-5         ($0.95 / $3.15 per M)
  kimi-k2.6, kimi-k2.5   from bedrock/moonshotai.kimi-k2.5  ($0.60 / $3.03)
  minimax-m2.7, m2.5     from bedrock/minimax.minimax-m2.5  ($0.36 / $1.44)
  deepseek-v4-pro, flash from azure_ai/deepseek-v3.2        ($0.58 / $1.68)
  qwen3.5/3.6-plus       from bedrock/qwen.qwen3-coder-next ($0.60 / $1.50)
  mimo-v2-pro/omni/2.5/2.5-pro from openrouter/xiaomi/mimo-v2-flash ($0.09 / $0.29)

Keys are prefixed `opencode/<slug>` to group with the rest of the
provider's catalog, mirroring how mistral/groq entries are stored.
The catalog's existing prefix-strip during lookup still resolves
the bare slug.

Cost reported is the upstream-equivalent value, not what the user
actually pays — OpenCode Go is a flat-rate $10/mo plan. Useful for
tracking the $60/mo cap consumption and comparing against direct
provider pricing. Users / docs should call this out as approximate.

Test plan: cargo test -p keplor-pricing — 12 passed.
@bravo1goingdark bravo1goingdark merged commit d6bb0eb into main May 5, 2026
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant