Search “DeepSeek”

activated per step, and supports text and image inputs. On Terminal-Bench 4.0, its score rose from 28% in the preview version to 41%, slightly above GLM-5.3's 40%, but on the new version of the Artificial Analysis Intelligence Index it scored only 26 points, below the 32 points of DeepSeek V4.1 Flash and GLM-5.3. API pricing is $0.70 per million input tokens and $2.10 per million output tokens, under a modified MIT license; using it as a commercial cloud service requires separate authorization.

Matched fieldsSummary、Body

orld's best open-weight model in automated coding and cybersecurity, and making it freely available. Pierre Stock, Mistral's VP of Science, said the model was trained on increasingly difficult cybersecurity and coding problems, enabling it to match or surpass models from Chinese labs such as DeepSeek and Z.ai. Stock also stressed that the West needs its own advanced open-weight models, because closed models can be shut down remotely, whereas open weights cannot have access cut off once released.

Matched fieldsSummary、Body

enterprise agents, responsible for providing production inference and post-training support for agents built on Postman Agent Stack. Baseten says its seven model APIs released in 2026 all ranked in the top two in Artificial Analysis's throughput and latency rankings, with Beam 1.0, DeepSeek V4.1 Flash and others taking first place. Harvey used Baseten to post-train a 27B open-source model, matching closed-source frontier models on legal agent benchmarks, while the latter cost about $50 per task.

Matched fieldsSummary、Body

Prime Intellect has released Prime Mega MoE, which builds on the Mega MoE kernel from DeepSeek's open-source DeepGEMM library by adding a fused and overlapped backward pass, making it usable for training rather than inference only. In the prime-rl trainer, the full training step for Qwen3 MoE models is 32% to 34% faster than the baseline and 48% to 55% faster than TorchTitan; the single-layer forward pass is up to 3.11x faster (from 12.0 ms to 3.9 ms), but the backward pass for Qwen3-235B-A22B i

Matched fieldsSummary、Body

6.1 Sol (max) at 55 points, with Gemini 4 Argon (high) at 50 points. The gap between open-weight models and the leaders has widened from 3 points to 12 points; the released-weight DeepSeek V4.1 Flash 1017 (max) and GLM-5.3 (max) both score 32 points. The per-task cost of leading models differs by more than 13 times: GPT-6.1 Terra (max) is $0.96, while Claude Fable 5.1 (max) is $13.04. A Progress Index v1 was also released, placing models released since November 2022 on the same capability scale

Matched fieldsSummary、Body

DeepSeek founder Liang Wenfeng, with a net worth of 280 billion yuan, debuted on the Hurun Rich List at No. 7 nationwide. The two logics he emphasized at a closed-door meeting were: proactively open-sourcing core capabilities so the entire industry grows on his foundation, and, amid the boom of wrapper monetization at the application layer, restraining short-term gains and relentlessly focusing on infrastructure. Sources say he has recently been intensively catching up on embodied intelligence i

Matched fieldsSummary

gents to execute, covering product selection, building, launch, and first-month profitability. It has two layers: a human-facing playbook/00–13 chapters and an Agent-facing AGENTS.md (Phase 0–14, with each step marked [AGENT] or [HUMAN]); before starting, one must pass the Gate A scoring. The tech stack is fixed on Cloudflare Workers, D1, Better Auth, OpenRouter routing to DeepSeek, Superwall plus Stripe, etc., and explicitly lists a long set of non-choices such as Vercel, Supabase, and Next.js.

Matched fieldsSummary、Body

-scoring Opus 5.5 scores only 45%, showing the benchmark is far from saturated. Compared with version 1.1, reference solutions modify on average 67% more files and 89% more lines of code, the codebase size is 25% larger, task-covered repositories increase by 55%, and blocking scoring criteria increase by 43%. DeepSeek V4.1 Flash drops from 43.9% to 19.3%, and Kimi K3 drops from 44.2% to 17.7%; open-source models, especially Western open-source models, still have significant room for improvement.

Matched fieldsSummary、Body

cing cut its costs by 80% to 90%. A U.S. bank with more than $7 billion in assets doubled its ChatGPT seats to 500 for free, paying less than $7 per user instead of the usual $20 or more. OpenAI is partly sacrificing its own profits to gain market share, but it has also achieved real price cuts through its Sol and Luna model lines; according to an investor memo, Anthropic's growth slowed from Q2 to Q3, and both companies also face low-price pressure from Chinese models such as DeepSeek and Kimi.

Matched fieldsSummary、Body

After having Doubao, Kimi, DeepSeek, GPT, Qwen and Claude each write 5 episodes of a restaurant business-war short drama, the 30 scripts were highly homogenized: noodle shops appeared 6 times and all six models chose a noodle shop for their first version, "conscience" appeared in every script, and "rules" in about 80%. Claude's dialogue is short and rhythmic, with an apprentice urging a price cut in every version, making it closest to the "heartburn style"; Doubao packs in punchy lines but 3 ver

Matched fieldsSummary、Body

oads since its September launch, of which 10.2 million came from the United States, making it one of the mobile apps to surpass 10 million U.S. downloads within 40 days of launch. DeepSeek open-sourced its inference framework DeepInfer and disclosed that a single cluster serves a throughput of 155 trillion tokens per day, with revenue of $38 per card per hour. Anthropic launched Claude Duet, a personal agent based on Claude Opus 5.5, opening research preview waitlist applications to Claude Pro a

Matched fieldsSummary、Body

.1 Pro (Max) ranks 11th in Arena's Agent Arena, achieving a +7.6% net improvement based on 4K+ real agent sessions, up 6 percentage points from Kimi K3 (Max)'s 25th place. Its median cost per task is $1.03, 13% lower than the previous version, ranking 8th, 9th, and 10th in the Work, Code, and Chat categories respectively. Arena stated that the model's open weights will be released soon, at which point it will become the top-ranked open-source model in Agent Arena, surpassing DeepSeek-v4.1-Flash.

Matched fieldsSummary、Body

model. The course first builds a math answer verifier and evaluates on MATH-500, then implements temperature sampling, top-p filtering, and self-consistency, followed by RLVR training with a simplified version of GRPO, and finally distills a small model using answers generated by DeepSeek R1. The content accompanies his book "Build a Reasoning Model (From Scratch)"; the early text generation and evaluation examples can run on a CPU, while later training experiments are recommended to use a GPU.

Matched fieldsSummary、Body

token consumption was about 500 trillion, nearly 4 times that of December 2025. In terms of performance, in August the average TPS of domestic models reached 78.23, doubling the 2025 average of 38.25, but the average TPS of typical foreign models was 112.82; the domestic first-token latency rose to 1.26 seconds. The report also compared four open-source models, DeepSeek, Qwen, GLM and Kimi, and found that the TPS gap for the same model across different platforms can be close to a factor of two.

Matched fieldsSummary、Body

Entertainment Capital had Doubao, Kimi, DeepSeek, GPT, Qwen and Claude each write 5 episodes of a restaurant business-war short drama, 30 scripts in total. Claude was deemed to have the most "Shaoxin flavor," with dialogue rhythm and functional roles such as apprentices and disciples closest to existing Shaoxin short dramas. Doubao came second, more like a punchline generator, but three versions of its scripts never specified what the shop sells. Across the 30 scripts, "conscience" appeared in a

Matched fieldsSummary、Body

s-5 (Max) at 70% lower cost; compared with Kimi K3 (Max)'s 17th place and 1655 points, it improves 9 places and 37 points. Kimi thus returns to the top 4 in the Code Arena: WebDev lab rankings. The model is priced at $2/$10 per million input/output tokens, with a median cost per task of $1.23, 29% lower than Kimi K3; the official announcement says open weights will be released soon, at which point it will become the top-ranked open-source model on the leaderboard, surpassing DeepSeek-v4.1-Flash.

Matched fieldsSummary

Kimi K3.1 Pro scores 64% on the Artificial Analysis Cyber Index, 7 points higher than the previous leader among openly available models, DeepSeek V4.1 Flash 1017 (57%), and 5 points lower than GPT-6 Sol (Daybreak Blue, max, 69%), which is available only through the OpenAI trusted access program. The index consists of three benchmarks: CWE-Bench-AA, DeepsecBench-AA, and CyberGym-E2E-AA. K3.1 Pro achieves 95% on CyberGym-E2E-AA, higher than GPT-6 Sol (Daybreak Blue)'s 93%. Its per-task cost is $2.

Matched fieldsSummary、Body

Matched fieldsSummary、Body

36Kr sorted through the product moves of eight companies—Tencent, Alibaba, Baidu, ByteDance, and Moonshot AI, Zhipu, MiniMax, DeepSeek—across ten AI tracks over four years, finding that competition cycles have shortened, product lifecycles have grown shorter, and renaming, integration, and shutdowns have become the norm. Tencent QClaw went from public beta on March 20 to announcing service shutdown arrangements on September 24, just 188 days, with user data to be migrated to WorkBuddy; Alibaba's

Matched fieldsSummary、Body

r solving robot AI, electromechanical design, reliability and lifespan issues by the end of next year, Optimus can be delivered to companies other than Tesla; Tesla will also build an Optimus Academy, planning to mass-produce 20,000 units for autonomous training. The domestic construction robot market reached 18.6 billion yuan in 2025 and is expected to exceed 80 billion yuan by 2030. DeepSeek is reported to have completed about 100 billion yuan in financing, with CATL and Tencent participating.

Matched fieldsSummary、Body