←Search results

Search “Reasoning”

By relevance · Total 206 results Clear search

Sunday

  6. Story1 reports · 0 facts

    Anthropic releases Claude Opus 5.5

    On September 22, 2026, Anthropic released Claude Opus 5.5. According to official materials, the model ranks first on Artificial Analysis's Intelligence Index v4.3 with a score of 58, 7 points higher than Claude Opus 5, and tops Vals AI's Vals Index at 69.69%; it supports up to 1 million tokens of input and 128,000 tokens of output, with reasoning enabled by default across five levels, and API pricing at $4/$0.25/$20 per million input/cached/output tokens. Earlier reports said its output price was $20 per million tokens, lower than the previous generation's $25. On September 23, Yuchen Jin said it was about 30% faster and about 40% cheaper per task than Opus 5. On September 25, Thomas Wolf said he used Opus 5.5 to generate a video from a Google Doc. On October 4, Ado Kukic said that using Opus 5.5 at medium and high effort he did not hit the 5-hour limit, using at most about 74% in a single session. On October 9, Ado again said that the weekend after Opus 5.5's release was his most productive coding day of the year so far. During the same period, Anthropic also released Sonnet 5.5 and Haiku 5.5, and lowered Sonn

    On September 22, 2026, Anthropic released Claude Opus 5.5. According to official materials, the model ranks first on Artificial Analysis's Intelligence Index v4.3 with a score of 58, 7 points higher than Claude Opus 5, and tops Vals AI's Vals Index at 69.69%; it supports up to 1 million tokens of input and 128,000 tokens of output, with reasoning enabled by default across five levels, and API pricing at $4/$0.25/$20 per million input/cached/output tokens. Earlier reports said its output price wa

    Matched fieldsSummary

  7. Story1 reports · 1 facts

    Fireworks evaluates DeepSeek's new model

    On September 14, 2026, Fireworks AI released and evaluated DeepSeek-V4.1-Flash. According to its published results, the model achieves 74.34% pass@1 on DeepSWE, on par with GPT-6-Astra's 74.12%; its cost per task is $0.43, about 1/15 that of Astra. The model is a 552B-parameter MoE, using separate activation budgets of 8B for input and 16B for output, and its KV Cache HBM footprint is reduced to 1/4 that of the previous-generation V4. On HLE it scores 34.52%, below Astra's 50.40%, and Fireworks considers it better suited to agentic coding than to highly difficult reasoning. On September 28, Fireworks AI released its first cache-aware router, FireRouter with Opus, which is now live in the CLI and offered for the first time as a standalone routing model; it can select among Claude Opus 5.5, GLM 5.3 and GLM 5.3 Flash based on task fit, prompt cache cost and switching cost; its internal A/B testing over more than a month showed coding task cost per session dropping from $15.36 to $6.63, a 57% reduction (±19 percentage points), with accuracy at pure O

    On September 14, 2026, Fireworks AI released and evaluated DeepSeek-V4.1-Flash. According to its published results, the model achieves 74.34% pass@1 on DeepSWE, on par with GPT-6-Astra's 74.12%; its cost per task is $0.43, about 1/15 that of Astra. The model is a 552B-parameter MoE, using separate activation budgets of 8B for input and 16B for output, and its KV Cache HBM footprint is reduced to 1/4 that of the previous-generation V4. On HLE it scores 34.52%, below Astra's 50.40%, and Fireworks

    Matched fieldsSummary

Tuesday

Wednesday

Sunday

  1. Story1 reports · 1 facts

    Ant's ASystem team launches AReno local post-training toolkit

    On August 14, 2026, Ant's ASystem team launched the AReno local large-model post-training toolkit. On the same day, Ant Ling released a practice article describing how AReno was used on DGX Spark to perform Agentic RL on Ling-3.0-tiny, using tic-tac-toe as the minimal validation task and completing a four-step closed loop of baseline evaluation, reward definition, post-training, and re-testing in the same environment. The experiment used GSPO to train for 400 steps; rollout/rewards_mean rose from about -0.5 to around 0.4, response_len dropped to about 850 tokens, and in the re-test, tool calls and move selection were more stable. Ling-3.0-tiny is a hybrid reasoning model with 7.9B total parameters and 1.3B activated parameters, and BF16, FP8, and INT4 versions have been open-sourced; AReno provides a Docker reproduction tutorial. The toolkit has now been made public together with the above practice.

    On August 14, 2026, Ant's ASystem team launched the AReno local large-model post-training toolkit. On the same day, Ant Ling released a practice article describing how AReno was used on DGX Spark to perform Agentic RL on Ling-3.0-tiny, using tic-tac-toe as the minimal validation task and completing a four-step closed loop of baseline evaluation, reward definition, post-training, and re-testing in the same environment. The experiment used GSPO to train for 400 steps; rollout/rewards mean rose fro

    Matched fieldsSummary

Friday