SA
Shanghai AI Laboratory
Multimodal model · China ·May 2026

Intern-S2-Preview

Multimodal Open Weights
35B
Parameters
262.1K
Context
2
Benchmarks

About Intern-S2-Preview

Intern-S2-Preview is an AI model developed by Shanghai AI Laboratory, in the multimodal category, released in 2026, made available as an open-weights model with 35B params. An efficient 35B scientific multimodal foundation model continued-pretrained from Qwen3.5, with image-text understanding.

On this page you'll find Intern-S2-Preview's full specifications, including a context window of 262.1K tokens, 2 benchmark results. Review provider pricing and benchmark scores below, or compare Intern-S2-Preview head-to-head with other multimodal models.

→ More models by Shanghai AI Laboratory → All multimodal models → AI benchmarks leaderboard

Run Intern-S2-Preview locally — what to buy

Share this build ↗

Sized to what Intern-S2-Preview actually needs (~262k context at Q4), not to the biggest GPU: Good = cheapest that runs it, Better = best value, Best = most headroom. Speed is a hardware estimate; anything past ~120 tok/s is shown as instant.

Good
AMD Radeon RX 7900 XTX
Consumer GPU · ~$999
~37.7 tok/s
Q4 · tight on context · up to 4k ctx
Runs with: ExLlamaV2, KoboldCpp, LM Studio
See full estimate →
Better
NVIDIA GeForce RTX 4090
Consumer GPU · ~$1,599
~44.7 tok/s
Q4 · tight on context · up to 4k ctx
Runs with: ExLlamaV2, KoboldCpp, LM Studio
See full estimate →
Best
NVIDIA GeForce RTX 5090
Consumer GPU · ~$1,999
~79.5 tok/s
Q4 · fits full context · up to 32k ctx
Runs with: ExLlamaV2, KoboldCpp, LM Studio
See full estimate →

Software support

·ExLlamaV2 ·KoboldCpp ·LM Studio ·MLX ·Ollama ·SGLang ·TensorRT-LLM ·llama.cpp ·vLLM ?KTransformers ?text-generation-webui

✓ measured · · compatible · — not supported. Informational only — speed is hardware-based.

Consumer GPU
10.1–79.5 tok/s
9 options →
Apple
12.6–33.5 tok/s
17 options →
Datacenter GPU
33.9–120+ tok/s
13 options →

Community

Would you run Intern-S2-Preview again?
No votes yet — be the first.
Did it run? — community reports by hardware

No reports yet. Own this model on some hardware? Be the first to confirm it runs.

Discussion

No comments yet. Start the discussion.

Intern-S2-Preview benchmark scores▶
Compare on the benchmark table →
Benchmark Category Score Variant
SWE-bench Verified code 64%
MMLU-Pro general 88%

More models like Intern-S2-Preview

AI
NeuroChess
· 72.3K
 SA
ETCHR
Shanghai AI Laboratory · 9B
 SA
InternLM3-8B-Instruct
Shanghai AI Laboratory
 Mistral AI
Mistral OCR 4
Mistral AI
 Tencent Hunyuan
Hy-Embodied-0.5-VLA
Tencent Hunyuan
 Anthropic
Claude Mythos 5
Anthropic

Frequently asked questions

What is Intern-S2-Preview? ▶

An efficient 35B scientific multimodal foundation model continued-pretrained from Qwen3.5, with image-text understanding It is tracked on GenAIList with its specifications, benchmark scores and provider pricing.

Who created Intern-S2-Preview? ▶

Intern-S2-Preview was developed by Shanghai AI Laboratory and released in 2026.

Is Intern-S2-Preview open source or proprietary? ▶

Intern-S2-Preview ships as an open-weights model: you can download and self-host the weights, though the licence may place some restrictions on use.

What is Intern-S2-Preview's context window? ▶

Intern-S2-Preview supports a context window of 262.1K tokens, and has 35B params.

How much does Intern-S2-Preview cost? ▶

Pricing for Intern-S2-Preview depends on the provider. See the providers table on this page for the latest API rates.

How does Intern-S2-Preview perform on benchmarks? ▶

Intern-S2-Preview is benchmarked across 2 evaluations on GenAIList, including SWE-bench Verified (64%). See the full benchmark table below and compare it with other models.

Intern-S2-Preview FP8: 35B Scientific Multimodal Model Running Locally

Know a great video about Intern-S2-Preview?