Best suited for
- 01
Streaming chat
Display tokens as they arrive to keep conversational interfaces responsive throughout generation.
- 02
Voice-agent reasoning
Use the model between transcription and speech synthesis to generate the next turn in a conversation.
- 03
Code-mixed dialogue
Handle conversations that move between Indian languages, romanised text and English.
Benchmarks
Evaluate the conversational variant on the work it will do.
A separate Chat evaluation is not published on Artificial Analysis.
Sarvam 105B Chat is distinct from the flagship reasoning model. The flagship comparison page also has no published quality scores. Compare Chat on your conversations, languages and response-time targets.
See the flagship model pageBuild with the Sarvam SDK
pip install --upgrade sarvamaiV1 · Sarvam SDK · Streamingimport osfrom sarvamai import SarvamAIclient = SarvamAI(api_subscription_key=os.environ["SARVAM_API_KEY"])stream = client.chat.completions(model="sarvam-105b-conversations",messages=[{"role": "user", "content": "Explain how solar panels work in Hindi briefly."}],max_tokens=8000,reasoning_effort=None,stream=True,)for chunk in stream:if not chunk.choices:continuetext = chunk.choices[0].delta.contentif text:print(text, end="", flush=True)
- 01 Install the Sarvam SDK
Install
sarvamaifor Python or JavaScript. Store your key inSARVAM_API_KEY. - 02 Choose your model
Use model ID
sarvam-105b-conversations. The SDK callshttps://api.sarvam.ai/v1for this example.V1 Chat Completions. Thinking is disabled for this short reply.
- 03 Validate your application
Check output quality, supported parameters and tool behaviour on representative requests before moving production traffic.
Frequently Asked
Questions
Explore other models
View catalogueSarvam 105B
- Sarvam
- Indian languages
- 128K context
Flagship 105B Mixture-of-Experts model trained from scratch for Indian-language understanding, complex reasoning, coding and agentic workflows.
GLM 5.3 Flash
- Z.ai
- Reasoning and coding
- 1M context
Faster open-weight reasoning model for agents and image understanding.
Gemma 4 31B
- Google DeepMind
- Vision
- 256K context
31B-parameter instruction-tuned model for text and image understanding, with tool calling and optional reasoning.
From model choice to production. Bring your workload, traffic expectations and latency targets. Build on Sarvam’s serving stack in India.
From model choice to production.
Bring your workload, traffic expectations and latency targets. Build on Sarvam’s serving stack in India.