SenseNova 商汤日日新 (SenseTime)
SenseTime's MoE multimodal models, from cloud API to edge boxes
Overview
SenseNova (日日新) is SenseTime's foundation-model line, led by the 600B-parameter MoE model V6.5 Pro, sold as token-metered APIs, an edge-deployable enterprise appliance, and open-sourced multimodal/spatial-intelligence models on GitHub.
What you can produce with SenseNova 商汤日日新 (SenseTime)
- SenseNova V6.5 Pro multimodal MoE API
- Open-sourced SenseNova-SI spatial-intelligence model
- Open-sourced SenseNova U1 unified image/text model
- SenseFoundry enterprise deployment platform
- Edge-deployable inference appliance
- Token Plan API subscription (free beta)
ASEAN Perspective
SenseNova 商汤日日新 (SenseTime) in Southeast Asia
ASEAN-region availability and pricing notes coming soon. Drop the editorial team a note via /contact/ if you can supply local context (Singapore/Malaysia/Indonesia/Thailand/Vietnam).
SenseTime's vision-first pedigree shows: SenseNova V6.5 Pro reportedly topped a Chinese multimodal benchmark ranking in 2025, and the company's open-sourced SenseNova-SI spatial-intelligence model claims to beat GPT-5 on spatial-reasoning tasks specifically, which is a narrower but credible claim given SenseTime's computer-vision roots. Token pricing is competitive on paper - V6 Pro API access has been quoted around ¥0.58 per million tokens - but the May 2026 Token Plan subscription is still in free public beta, with paid Lite and Pro tiers announced but not yet priced. The enterprise story is real: SenseFoundry has shipped into finance, government and healthcare accounts, and an edge appliance claims roughly 80% lower inference cost than comparable on-prem hardware. What's thin is anything resembling a self-serve consumer product - SenseChat has minimal presence next to Doubao or Kimi - so this is squarely an enterprise/API buy, sold through sales rather than a credit card.
What people say
SenseNova is SenseTime's model line, built on the company's two-decade run in computer vision rather than a from-scratch LLM play. The current flagship, V6.5 Pro, is a native mixture-of-experts multimodal model with long chain-of-thought reasoning; SenseTime's own benchmark writeup put it at the top of a 2025 Chinese ranking with a score of 75.35, ahead of Doubao-Seed-1.6-vision, Ernie 5.0 and Qwen3 on the metrics it cites - worth taking with the usual grain of salt that comes with vendor-reported benchmarks. More interesting is SenseNova-SI, an open-sourced spatial-intelligence model that SenseTime says outperforms GPT-5 and Gemini on spatial-reasoning tasks specifically, and SenseNova U1, a unified image-and-text model built on what the company calls a NEO-unify architecture that skips the usual VAE step in image generation - both are published on the OpenSenseNova GitHub org.
On pricing, the numbers are scattered across models: V6 Pro API access has been quoted at roughly ¥0.5833 per million tokens, while V6.5 Pro runs closer to ¥9 per million tokens. In May 2026 SenseTime launched a 'Token Plan' subscription - currently a free public beta offering 1,500 calls per 5-hour window across three models (SenseNova 6.7 Flash-Lite, SenseNova U1 Fast, and a hosted DeepSeek V4 Flash) - with paid Lite and Pro tiers announced for later rollout but not yet priced anywhere public.
On the enterprise side, SenseFoundry is the deployment platform behind named accounts: SenseTime has run a large-model partnership with Kingsoft's WPS 365 since 2023, and co-published a financial-services large model with Haitong Securities covering compliance, risk control and customer service. A separate edge appliance aimed at finance, coding, healthcare and government use claims about an 80% inference-cost reduction versus comparable on-prem hardware. What's largely absent is independent third-party review coverage - no G2/Capterra listings turned up, and public sentiment is mostly Chinese tech-press coverage of product launches rather than user reviews, so buyers are relying on SenseTime's own benchmark claims more than outside verification.
Summary of public user & expert reviews, compiled by RECATOOLS.
About this listing
This entry was compiled from publicly available data including SenseNova 商汤日日新 (SenseTime)'s official website, press releases, documentation, and reputable third-party publications. RECATOOLS is not affiliated with SenseNova 商汤日日新 (SenseTime) unless explicitly stated.
Third-party AI tools update their pricing, features, availability, and policies frequently. Information here may be outdated by the time you read this — we make reasonable efforts to keep listings current, but cannot guarantee absolute accuracy.
For the latest details, please refer to SenseNova 商汤日日新 (SenseTime) directly →
Spotted something out of date? Suggest an update →
Alternatives to SenseNova 商汤日日新 (SenseTime)
More in LLMs & Chat
