Sarvam-105B

by Sarvam AIChat

Flagship Indian-language reasoning model for enterprise multilingual applications

Context
131K tokens
Max output
131,072 tokens
Released
2026-02-18
Weights
Open

sarvam/sarvam-105b

Specifications

Full facts as recorded, incl. what is unknown.

Released
2026-02-18
Weights
Open
Input
Text
Output
Text
Context window
131,072 max across endpoints
Max output
131,072
Providers
1 + 2 gateways, not counted as providers
Epoch Capabilities Index
not evaluated
Direct price range
$0.04 in at FastRouter, via gateway, fees not verified
More details
Last updated
2026-03-06
Knowledge cutoff

Capabilities

Across all endpoints: supported if any endpoint supports it.

  • Tool calling
  • Parallel toolsunknown
  • Structured output
  • Reasoning
  • Image input
  • PDF input
  • Audio input
  • Audio outputunknown
  • Video input
  • Prompt cachingunknown
  • Web searchunknown
  • Computer useunknown
  • System messagesunknown
  • Assistant prefillunknown
  • Temperature
  • Streamingunknown
  • Attachments

Endpoints

One row per provider and variant; gateways are listed separately.

Loading endpoints…

Cost calculator

Loading endpoints…

Similarly priced models

Blended rate per 1M tokens, each at its own cheapest provider (10M input : 2M output mix).

ModelBlended/1MIn/1MOut/1MContextvs Sarvam-105B
Seed 1.6 FlashByteDance Seed$0.0617$0.03$0.22256K+3%
Mercury 2.5 PreviewInception$0.0583$0.04$0.15260K−3% cheaper
Command R7BCohere$0.0563$0.0375$0.15128K−6% cheaper
Llama 3.2 1b InstructMeta$0.056$0.027$0.201131K−7% cheaper
Llama 3.1 8B InstructMeta$0.055$0.05$0.08200K−8% cheaper
llama-3-8bMeta$0.055$0.05$0.088.2K−8% cheaper
Mistral Small 3Mistral AI$0.055$0.05$0.0832.8K−8% cheaper
Phi-4Microsoft$0.0653$0.056$0.11216.4K+9%

More from Sarvam AI

Newest Sarvam AI models in the catalogue. All Sarvam AI models

ModelReleasedContextListed price
Sarvam-30B2026-02-1865.5K$0.02 via FastRouter
sarvam-m2025-07-258.2Kunpriced
Heads up: prices are indicative list rates scraped from provider pricing pages; verify with the provider before relying on them. Report incorrect data for this model