Summary
Together AI serves pearl-ai/gemma-4-31b-it on their public serverless inference surface. This is Pearl Research Labs' INT8-quantized instruction-tuned checkpoint of Google's Gemma 4 31B, offered at lower pricing than Google's own google/gemma-4-31B-it ($0.28/$0.86 vs $0.39/$0.97 per MTok) but with a smaller 32K context window (vs 262K).
The model ID pearl-ai/gemma-4-31b-it is entirely absent from model_list.json.
Missing Model
| Model ID |
Provider |
Input $/MTok |
Output $/MTok |
Context |
Status |
pearl-ai/gemma-4-31b-it |
together |
$0.28 |
$0.86 |
32,000 |
Active (serverless) |
Verified Specifications
| Field |
Value |
Source |
| format |
"openai" |
Together uses OpenAI-compatible API; matches all other Together entries in catalog |
| flavor |
"chat" |
Listed in Chat models table on serverless docs |
| input_cost_per_mil_tokens |
0.28 |
Together pricing page |
| output_cost_per_mil_tokens |
0.86 |
Together pricing page |
| displayName |
"Gemma 4 31B IT (Pearl)" |
Derived from model card title "Gemma-4-31B-it-Pearl" |
| max_input_tokens |
32000 |
Together serverless models page — context length column |
| available_providers |
["together"] |
Only served by Together AI |
Fields NOT published
- max_output_tokens — not documented on Together's serverless models page or pricing page; omitted rather than guessed
- multimodal — not stated; the model is listed as text-only input/output on the serverless models page
Comparison with Google's official Gemma 4 31B on Together
| Attribute |
google/gemma-4-31B-it |
pearl-ai/gemma-4-31b-it |
| Context Length |
262,144 |
32,000 |
| Input Price |
$0.39/M |
$0.28/M |
| Output Price |
$0.97/M |
$0.86/M |
| Function Calling |
Yes |
No |
| Structured Outputs |
Yes |
No |
Verification Checklist
| Check |
Status |
Detail |
| Cross-source |
✅ |
Listed on both Together serverless models page (model listing with context and capabilities) AND Together pricing page (pricing table) |
| Already fixed |
✅ |
No pearl-ai entries exist anywhere in model_list.json |
| ID format |
✅ |
Follows Together's org/model naming convention (e.g., deepcogito/cogito-v2-1-671b, essentialai/rnj-1-instruct, LiquidAI/LFM2-24B-A2B) |
| Duplicate check |
✅ |
No open issue covers pearl-ai/gemma-4-31b-it. Issue #675 mentions it in passing but targets google/gemma-4-31b-it — a different model ID with different specs |
Local Files Inspected
packages/proxy/schema/model_list.json — no pearl-ai entries foundpackages/proxy/schema/index.ts — no pearl-ai entries found
{
"kind": "missing_model",
"provider": "together",
"models": ["pearl-ai/gemma-4-31b-it"],
"status": "active",
"model_specs": {
"pearl-ai/gemma-4-31b-it": {
"format": "openai",
"flavor": "chat",
"input_cost_per_mil_tokens": 0.28,
"output_cost_per_mil_tokens": 0.86,
"displayName": "Gemma 4 31B IT (Pearl)",
"max_input_tokens": 32000,
"available_providers": ["together"]
}
},
"source_urls": [
"https://docs.together.ai/docs/serverless-models",
"https://www.together.ai/pricing"
]
}
Summary
Together AI serves
pearl-ai/gemma-4-31b-iton their public serverless inference surface. This is Pearl Research Labs' INT8-quantized instruction-tuned checkpoint of Google's Gemma 4 31B, offered at lower pricing than Google's owngoogle/gemma-4-31B-it($0.28/$0.86 vs $0.39/$0.97 per MTok) but with a smaller 32K context window (vs 262K).The model ID
pearl-ai/gemma-4-31b-itis entirely absent frommodel_list.json.Missing Model
pearl-ai/gemma-4-31b-itVerified Specifications
"openai""chat"0.280.86"Gemma 4 31B IT (Pearl)"32000["together"]Fields NOT published
Comparison with Google's official Gemma 4 31B on Together
google/gemma-4-31B-itpearl-ai/gemma-4-31b-itVerification Checklist
pearl-aientries exist anywhere inmodel_list.jsonorg/modelnaming convention (e.g.,deepcogito/cogito-v2-1-671b,essentialai/rnj-1-instruct,LiquidAI/LFM2-24B-A2B)pearl-ai/gemma-4-31b-it. Issue #675 mentions it in passing but targetsgoogle/gemma-4-31b-it— a different model ID with different specsLocal Files Inspected
packages/proxy/schema/model_list.json— nopearl-aientries foundpackages/proxy/schema/index.ts— nopearl-aientries found{ "kind": "missing_model", "provider": "together", "models": ["pearl-ai/gemma-4-31b-it"], "status": "active", "model_specs": { "pearl-ai/gemma-4-31b-it": { "format": "openai", "flavor": "chat", "input_cost_per_mil_tokens": 0.28, "output_cost_per_mil_tokens": 0.86, "displayName": "Gemma 4 31B IT (Pearl)", "max_input_tokens": 32000, "available_providers": ["together"] } }, "source_urls": [ "https://docs.together.ai/docs/serverless-models", "https://www.together.ai/pricing" ] }