Support · Sarvam-30B, Sarvam-105B and Sarvam-105B-1711

Will it last?

Ownership levelPartialnone·limited·partial·substantial·fullAnalytical input D ยท 44.4/100

This page is a projection of the one entry record, the Transparency factor that Support covers. The full verdict is set by all four factors together, floor-weighted so the weakest caps the whole.

Which domain expands which factor
  • AssessUse & modify + Transparency
  • ImplementData control + Reliability
  • UseReliability
  • SupportTransparency

Common problems & fixes

Reported in Hugging Face discussions and on the cards (ev-discussions, ev-card-105b, ev-card-1711):

  • vLLM does not load the March builds natively. Use the fork or hotpatch_vllm.py.
  • An is_torch_fx_available import error appears with newer Transformers. A staff fix for Transformers 5 loading and RoPE was merged on 12 November 2026.
  • Thinking cannot be turned off on some builds. Pass enable_thinking=False, or chat_template_kwargs on vLLM for 1711.
  • Short max_tokens values are consumed by reasoning. Raise the limit.

Versions, changelog & cadence

The 30B and 105B were created on 3 March 2026, with FP8 and GGUF builds in March. A fix for 105B MoE routing was merged on 21 September. A fix for 30B Transformers 5 loading was merged on 12 November. Sarvam-105B-1711 was released on 17 November, with a DSpark vLLM fix on 18 November. Revisions are Hugging Face commits (ev-hub, ev-card-1711).

The meaning of "1711" is not stated on the cards. The 105B discussion #14 asked for newer checkpoints on 29 October 2026 (ev-discussions).

Security / vulnerability disclosure

No security policy covers the weights or the custom code. The Trust Center lists a security email for the hosted platform, and its audit reports are shared under NDA (ev-trust).

Community & support channels

Hugging Face discussions on each repo, the sarvamai GitHub organisation, and enterprise support plans on the hosted platform (ev-discussions, ev-quant-code, ev-trust).

Sarvam also hosts the 105B on its own API and documents zero retention for Model APIs. Its binding Privacy Policy sets a 30-day default retention, and its documentation says retention is indefinite until set. The Privacy Policy states an opt-in default for training and then describes opt-out. The EULA promises opt-in only, and the Terms of Service control on conflict. Do not rely on the hosted API for confidential data before Sarvam resolves these (ev-privacy, ev-retention-docs, ev-eula, ev-tos).

Deprecation / end-of-life policy

No deprecation policy covers the weights. The hosted API has deprecated Sarvam-M and Sarvam-30B and tells users to migrate to Sarvam-105B (ev-api-docs).

Tracked known issues

Open Hugging Face threads cover loading errors, the thinking toggle, MLX support, newer checkpoints and training-data requests (ev-discussions). The blog and the 30B card disagree on the AIME 25 score. Artificial Analysis reports high hallucination for the March builds (ev-aa).

How this scores

The ownership factor this domain covers, drawn from the one entry record.

2

TransparencyDo you know what it is: weights, training, behaviour, and legible terms?

Moderate

Weights, cards and a blog are public. Training data, training code, post-training recipes, base checkpoints and a technical report are not (ev-card-105b, ev-blog, ev-no-report).

How this scores (AOI sub-dimensions)
Provenance2/5how well we can trace and verify what went into the modelThe canonical source is the sarvamai Hugging Face organisation, which is not verified.
Governance3/5how accountable and well-documented the publisher isSarvam AI is an accountable company in Bengaluru with a release record: the March release, fix pull requests in September and November, and the 1711 build in November.
What this means for adoptionYou can run and fine-tune Apache-2.0-declared weights, but you get no data, code, recipes or safety case, and the licence is a card declaration with no LICENSE file. Sarvam's own Terms and EULA restrict derivative works and training without mentioning the weights. Ownership rises if Sarvam adds a LICENSE file, clarifies those documents and publishes a safety evaluation.

Sources

The same evidence records as the entry sheet. Read means the text was verified; unverified means it is known to exist but not yet read.

Model cardread2026-11-19
The 105B card declares Apache-2.0 and lists 22 language codes plus English.
Model cardread2026-11-19
The 30B card declares Apache-2.0 and the same language codes.
Model cardread2026-11-19
The 105B-1711 cards declare Apache-2.0 and say the model shares the Sarvam-105B base and is further post-trained for tool calling, instruction following, agentic tasks and Indian languages.
Licenceread2026-11-19
All eight weights repos declare Apache-2.0 in the card front matter and none holds a LICENSE file (raw LICENSE returns HTTP 404).
Documentationread2026-11-19
The sarvamai Hugging Face organisation is not verified and lists 17 ungated model repos.
Vendor announcementread2026-11-19
The launch blog of 6 March 2026 says both models were trained from scratch on in-house data, 16T tokens for the 30B and 12T for the 105B, with the 10 most-spoken Indian languages emphasised in pre-training (publisher claims).
Documentationread2026-11-19
The indivibe dataset holds 4,840 evaluation prompts in the 22 scheduled languages, including Bodo, in native and romanised script, under CC-BY-4.0.
Documentationread2026-11-19
Sarvam's API documentation lists 10 Indian languages plus English for Sarvam-105B and for the deprecated Sarvam-30B.
Documentationread2026-11-19
The platform documentation says workspace data is retained indefinitely until an owner sets a period.
Terms of serviceread2026-11-19
Terms of Service v2.0, effective 29 July 2026, clause 10.5 bars using the Offerings or their output to train or fine-tune AI systems and bars discovering model weights.
Privacy Policyread2026-11-19
The Privacy Policy of 29 July 2026 headlines model training as opt-in but says Sarvam trains on inputs and outputs unless the user opts out.
Terms of serviceread2026-11-19
The EULA of 22 May 2026 defines Software to include 'proprietary language models' and bars derivative works.
Securityread2026-11-19
The Trust Center claims ISO 27001, SOC 2 Type II, India-only residency and no use of one customer's data to train models for other customers.
Third-party analysisread2026-11-19
Artificial Analysis scored Sarvam 105B at 18 and Sarvam 30B at 12 on its Intelligence Index in April 2026.
Documentationread2026-11-19
Official FP8 builds of both models and Q4_K_M GGUF builds (19.6 GB for the 30B, 64.2 GB for the 105B) exist.
Documentationread2026-11-19
Community threads ask for the training data and for newer checkpoints and have no Sarvam reply.
Vendor announcementread2026-11-19
The Token Factory post of 16 November 2026 says Sarvam serves the 105B in MXFP8 with DSpark speculative decoding from GPU clusters in India.
Technical_reportunverified2026-11-19
No technical report, paper or data statement for these models was found.
Documentationunverified2026-11-19
The AIKosh page for the 105B, linked from the blog as a second download source, is a JavaScript app and returns no text.