Gap
Three models in packages/proxy/schema/model_list.json currently list only baseten as their available provider, but all three are also available on Together AI's serverless inference platform with published pricing.
Affected Entries
| Model ID |
Current providers |
Together Input $/MTok |
Together Output $/MTok |
zai-org/GLM-5 |
["baseten"] |
$1.00 |
$3.20 |
moonshotai/Kimi-K2.5 |
["baseten"] |
$0.50 |
$2.80 |
moonshotai/Kimi-K2.6 |
["baseten"] |
$1.20 |
$4.50 |
Note: zai-org/GLM-5.1 already has ["together"] in its available_providers, so this is an inconsistency — GLM-5 from the same family is missing Together.
Recommended Change
Add "together" to the available_providers array for each of these three models. No new model entries are needed — only the provider list needs updating.
Verification
Both sources are official Together AI documentation confirming these model IDs are available on the platform.
Fields NOT changed
- Pricing fields — the catalog pricing reflects baseten rates. Together's rates differ slightly (e.g., GLM-5: baseten $0.95/$3.15 vs Together $1.00/$3.20). The catalog does not currently store per-provider pricing, so this is noted but not actionable.
moonshotai/Kimi-K2.6 has no pricing in the catalog currently. Together lists $1.20/$4.50 — the downstream fix job may want to add pricing.
Local files inspected
packages/proxy/schema/model_list.json — confirmed zai-org/GLM-5 (line ~9752), moonshotai/Kimi-K2.5 (line ~5473), moonshotai/Kimi-K2.6 (line ~5482) all have available_providers: ["baseten"]
packages/proxy/schema/model_list.json — confirmed zai-org/GLM-5.1 (line ~9804) already has available_providers: ["together"]
Source URLs
{
"kind": "provider_mapping",
"provider": "together",
"models": ["zai-org/GLM-5", "moonshotai/Kimi-K2.5", "moonshotai/Kimi-K2.6"],
"status": "active",
"model_specs": {
"zai-org/GLM-5": {
"available_providers": ["baseten", "together"]
},
"moonshotai/Kimi-K2.5": {
"available_providers": ["baseten", "together"]
},
"moonshotai/Kimi-K2.6": {
"available_providers": ["baseten", "together"]
}
},
"source_urls": [
"https://docs.together.ai/docs/serverless-models",
"https://docs.together.ai/docs/glm-5-quickstart"
]
}
Gap
Three models in
packages/proxy/schema/model_list.jsoncurrently list onlybasetenas their available provider, but all three are also available on Together AI's serverless inference platform with published pricing.Affected Entries
zai-org/GLM-5["baseten"]moonshotai/Kimi-K2.5["baseten"]moonshotai/Kimi-K2.6["baseten"]Note:
zai-org/GLM-5.1already has["together"]in itsavailable_providers, so this is an inconsistency — GLM-5 from the same family is missing Together.Recommended Change
Add
"together"to theavailable_providersarray for each of these three models. No new model entries are needed — only the provider list needs updating.Verification
zai-org/GLM-5,moonshotai/Kimi-K2.5,moonshotai/Kimi-K2.6) and per-token pricingzai-org/GLM-5as the model ID with code examples; Together AI Kimi K2.5 quickstart referenced in docs indexBoth sources are official Together AI documentation confirming these model IDs are available on the platform.
Fields NOT changed
moonshotai/Kimi-K2.6has no pricing in the catalog currently. Together lists $1.20/$4.50 — the downstream fix job may want to add pricing.Local files inspected
packages/proxy/schema/model_list.json— confirmedzai-org/GLM-5(line ~9752),moonshotai/Kimi-K2.5(line ~5473),moonshotai/Kimi-K2.6(line ~5482) all haveavailable_providers: ["baseten"]packages/proxy/schema/model_list.json— confirmedzai-org/GLM-5.1(line ~9804) already hasavailable_providers: ["together"]Source URLs
{ "kind": "provider_mapping", "provider": "together", "models": ["zai-org/GLM-5", "moonshotai/Kimi-K2.5", "moonshotai/Kimi-K2.6"], "status": "active", "model_specs": { "zai-org/GLM-5": { "available_providers": ["baseten", "together"] }, "moonshotai/Kimi-K2.5": { "available_providers": ["baseten", "together"] }, "moonshotai/Kimi-K2.6": { "available_providers": ["baseten", "together"] } }, "source_urls": [ "https://docs.together.ai/docs/serverless-models", "https://docs.together.ai/docs/glm-5-quickstart" ] }