baseten
Baseten: Mammoth 1
baseten/mammoth-1
Our take
AI-assisted analysis based on the sources below. Suitability is an editorial assessment; test it on your own workload.
baseten/mammoth-1 is a 27B-parameter decision model post-trained from Qwen3.5-27B by Base Labs, Baseten's research team. Rather than generating free text, it takes a state (text, optionally with images) plus typed questions — yes/no, multiple choice, or ordinal scores — and returns a calibrated probability for each option in a single shared-prefix pass. The vendor reports leading results on decision-focused benchmarks, though these were not independently verified in this source pack. It supports a 16,384-token context, is documented for English, and is released under an MIT license.[1][2]
Strengths
- Answers multiple typed questions from one shared-prefix prefill, returning a calibrated probability per option; training included a Brier calibration term and a trained "I don't know" option.[2]
- Accepts text and image inputs with a frozen vision encoder, and was fully fine-tuned on decision data rather than adapted via prompting.[1][2]
- The vendor reports top scores on two decision benchmarks, the Jev Decision Index 0.4 and a replica of the Artificial Analysis Decision Index — measured performance reported by the provider, not independently confirmed.[2]
Limitations
- Benchmark results are vendor-reported only; no independent evaluation was included in this source pack, so validate calibration and accuracy on your own workload.[1][2]
- Output is limited to decisions — it does not generate free-form text or reasoning, and the catalog lists no supported sampling parameters.[1]
- Only English is documented, no knowledge cutoff is reported, and requests exceeding the trained length are refused rather than truncated.[1][2]
Where to use it
- A candidate for routing and triage, such as assigning tickets to teams, matching the model card's classification and routing tags.[2]
- A candidate for guardrails and policy checks via yes/no questions, with a trained abstention option per the vendor description.[2]
- Ordinal scoring such as urgency levels over text or image-plus-text inputs; validate calibration on your own data first.[1][2]
Sources
- [1] OpenRouter catalog · Baseten: Mammoth 1Sources checked: Nov 13, 2026
- [2] Model documentation · Baseten: Mammoth 1Sources checked: Nov 13, 2026
About this model
Description below is the original text supplied by OpenRouter, shown in the source language.
Mammoth 1 is a 27B decision model from Base Labs, Baseten's research team, built on Qwen3.5-27B. Given a state (text or JSON) and typed questions (yes/no, multiple choice, or ordinal...
Specifications
- Model ID
- baseten/mammoth-1
- Canonical ID
- baseten/mammoth-1-20261021
- Creator
- baseten
- Added to OpenRouter
- 21.10.2026
- Context length
- 16,384
- Modality
- text+image->decisions
- Input modalities
- text, image
- Output modalities
- decisions
- Token pricing
- No per-token price in the source snapshot — see OpenRouter for the billing unit
- Knowledge cutoff
- —
- Expiration date
- —
Tasks and capabilities
A grade appears only where the source supplies a matching measurement for that task, the model is active and the cohort holds at least 10 comparable models. The capability column lists facts the source declares — supported formats, reasoning and tool parameters — which are not measured quality.
| Task | Grade | Source-declared capability |
|---|---|---|
| General quality | No measurement | — |
| Coding | No measurement | — |
| Agents | No measurement | — |
| Writing | No measurement | — |
| Reasoning | No measurement | — |
| Document analysis | No measurement | — |
| Vision | No measurement | input: image |
| Image generation | No measurement | — |
| Audio | No measurement | — |
| Video | No measurement | — |
| Embeddings | No measurement | — |
Evaluations
No evaluations are supplied for this model.
How grades work
Each letter grade compares a model only against other models scored by the same source and metric in the current snapshot. S means the top 10%, A the next 20%, B the next 30%, C the rest. Models without a published measurement are shown as not rated — a supported format alone never implies measured quality.
- S — S — top 10% of the cohort
- A — A — next 20% of the cohort
- B — B — next 30% of the cohort
- C — C — below the B band
Providers
Optional live provider statistics from OpenRouter. Missing values are unknown, not zero.
| Provider | Context length | Price in /1M | Price out /1M | Latency p50 (30m, s) | Throughput p50 (30m, tok/s) | Uptime (30m) |
|---|---|---|---|---|---|---|
| BaseTen | 16,384 | — | — | — | — | 100% |
Provider data retrieved: 13.11.2026
Related articles
No published articles identify this model yet.