English translation is being prepared.
Matched fieldsTitle
←Search results
Search “Model releases”
Qwen3.8-27B RL Post-Training Model Released
On October 20, 2026, an open-source environment model post-trained with reinforcement learning based on Qwen3.8-27B was released. According to Rohan Paul, the model reduced the hallucination rate on AA-Omniscience from 30.4% to 15.1%, and surpassed Muse Glimmer 30B and GPT-6 Luna high on Terminal-Bench, τ3-bench, ClawEval, and BFCL v4; however, on DeepSWE, Luna still leads with 53.1%. Its median output for solved tasks is 23.8K tokens, compared with 54.9K for the base Qwen. The authors also mention that the MLA version replaces 16 layers of full attention with multi-head latent attention, reducing memory usage per token from 64 KiB to 18 KiB; the compressed version can run on a single 24 GB RTX 4090, with a context about 3.5 times that of the base Qwen.
On October 20, 2026, an open-source environment model post-trained with reinforcement learning based on Qwen3.8-27B was released. According to Rohan Paul, the model reduced the hallucination rate on AA-Omniscience from 30.4% to 15.1%, and surpassed Muse Glimmer 30B and GPT-6 Luna high on Terminal-Bench, τ3-bench, ClawEval, and BFCL v4; however, on DeepSWE, Luna still leads with 53.1%. Its median output for solved tasks is 23.8K tokens, compared with 54.9K for the base Qwen. The authors also ment
Matched fieldsTitle
Qwen3.8-27B RL post-training model released
On October 20, 2026, an open-source environment model post-trained with reinforcement learning based on Qwen3.8-27B was released. According to Rohan Paul, the model reduced the hallucination rate on AA-Omniscience from 30.4% to 15.1%, and surpassed Muse Glimmer 30B and GPT-6 Luna high on Terminal-Bench, τ3-bench, ClawEval, and BFCL v4; however, on DeepSWE, Luna still leads with 53.1%. Its median output for solved tasks is 23.8K tokens, compared with 54.9K for the base Qwen. The authors also mention that the MLA version replaces 16 layers of full attention with multi-head latent attention, reducing memory usage per token from 64 KiB to 18 KiB; the compressed version can run on a single 24 GB RTX 4090, with a context about 3.5 times that of the base Qwen.
Matched fieldsTitle
Artificial Analysis launches HealthBench and HealthBench…
Artificial Analysis has launched leaderboards for OpenAI's two health dialogue benchmarks, HealthBench and HealthBench Professional; the former contains 5,000 conversations with general users and health workers, while the latter consists of 525 conversations all aimed at clinicians. On HealthBench Professional, GPT-6 Astra (max) ranks first with a length-adjusted score of 66.0%, GPT-6.1 Sol (max) is at 65.1%, and Claude Sonnet 5.5 (max with fallback) is at 63.0%. On HealthBench, Claude Sonnet 5.5 (max with fallback) leads at 62.1%, while DeepSeek V4.1 Flash (max) is at 60.9%; Claude Opus 5.5 and Claude Fable 5.1 average about 5,900 characters per answer and lose more than 11 points after length adjustment.
Artificial Analysis has launched leaderboards for OpenAI's two health dialogue benchmarks, HealthBench and HealthBench Professional; the former contains 5,000 conversations with general users and health workers, while the latter consists of 525 conversations all aimed at clinicians. On HealthBench Professional, GPT-6 Astra (max) ranks first with a length-adjusted score of 66.0%, GPT-6.1 Sol (max) is at 65.1%, and Claude Sonnet 5.5 (max with fallback) is at 63.0%. On HealthBench, Claude Sonnet 5.
Matched fieldsBody
Author trained six models in two days using ML Intern prompts
Hugging Face blogger yvrjsharma released seven prompts they used with ML Intern and showcased six projects completed within two days, including a citrus disease VLM based on Qwen3.5-2B, a Huggy character LoRA for FLUX.2 klein 4B, a camera viewpoint LoRA and a doodle LoRA for Qwen-Image 2.1, Pocket Rewriter 0.8B/2B distilled models, and a version of Agate Preview 002 distilled from 50 steps to 4 steps. The author recommends stating verified facts in prompts, requiring a pre-training baseline and smoke tests with checks, and setting spending caps, for example limiting the citrus project to 12 dollars. All prompts are available on GitHub at yvrjsharma/ml-intern-prompts, and ML Intern mode can be enabled in HuggingChat.
Hugging Face blogger yvrjsharma released seven prompts they used with ML Intern and showcased six projects completed within two days, including a citrus disease VLM based on Qwen3.5-2B, a Huggy character LoRA for FLUX.2 klein 4B, a camera viewpoint LoRA and a doodle LoRA for Qwen-Image 2.1, Pocket Rewriter 0.8B/2B distilled models, and a version of Agate Preview 002 distilled from 50 steps to 4 steps. The author recommends stating verified facts in prompts, requiring a pre-training baseline and
Matched fieldsBody
Fireworks Evaluates DeepSeek V4 Pro Against Fable 5 and Kimi K3
Fireworks tested DeepSeek V4 Pro 0813 on four benchmarks: SWE-Bench Verified, LiveCodeBench v6, Aider Polyglot, and Terminal-Bench 2.1. It scored 95.2% on SWE-Bench Verified, higher than Kimi K3's 92.6% and Fable 5's 85.4%, and 92.0% on LiveCodeBench v6. On cost, DeepSeek V4 Pro spent $0.309 per task solved on SWE-bench, versus $0.808 for Fable 5; on LiveCodeBench the figures were $0.040 and $0.225, respectively. Its weakness is the multilingual breadth of Aider Polyglot, where it scored only 48.9% on Java, versus 74.5% for Fable 5. After oracle routing on a combined set of 409 tasks, accuracy was 92.4%, higher than the 86.6% achieved by Fable 5 alone, at a cost of $0.279 versus $4.510 per task, with 86.3% of tasks routed to V4 Pro.
Fireworks tested DeepSeek V4 Pro 0813 on four benchmarks: SWE-Bench Verified, LiveCodeBench v6, Aider Polyglot, and Terminal-Bench 2.1. It scored 95.2% on SWE-Bench Verified, higher than Kimi K3's 92.6% and Fable 5's 85.4%, and 92.0% on LiveCodeBench v6. On cost, DeepSeek V4 Pro spent $0.309 per task solved on SWE-bench, versus $0.808 for Fable 5; on LiveCodeBench the figures were $0.040 and $0.225, respectively. Its weakness is the multilingual breadth of Aider Polyglot, where it scored only 48
Matched fieldsBody