Releases·Model Releasefeatured
Meta Releases Muse Spark 1.3, matching Fable 5
Meta released Muse Spark 1.3 on 2026-09-02, targeting improved performance on agentic and coding tasks and real-world usability. The model rolled out in Muse Code and the Meta Model API, with the Contributor variant available on OpenCode Go and a max-reasoning mode planned after safety testing. A Reddit post reports pricing of $0.10 input / $0.20 output per million tokens, matching Fable 5.
9322 engagement7 sourcesWed, Sep 2, 2026, 07:36 PM
Developments
Muse Spark 1.3 is now free on OpenCode as of 2026-09-03.
Thu, Sep 3, 2026, 02:05 AM
AI Overview
- A Reddit post announced the Muse Spark 1.3 release on 2026-09-02, drawing 89 upvotes
- Muse Spark 1.3 is priced at $0.10 per million input tokens and $0.20 per million output tokens, matching Fable 5, per Meta's developer page
- An x post from Meta on 2026-09-02 (2,518 engagements) says Spark 1.3 sustains longer-horizon agentic work across multiple workflows in a single thread and collaborates more actively by asking users questions
- Meta says max reasoning for Muse Spark 1.3 will follow in Muse Code and the Meta Model API after safety testing completes
- An x post on 2026-09-02 (2,221 engagements) says Meta Muse Spark 1.3 Contributor is now available on OpenCode Go
- An x post on 2026-09-03 (2,458 engagements) says Meta Muse Spark 1.3 is now free on OpenCode
Related
Social·Community
Muse Spark 1.3 (xhigh, standard tier) takes 2–3x longer than Claude to finish the same small task
A developer timed identical agent tasks building and deploying a static site to Cloudflare Workers, comparing Meta's muse-spark-1.3 (xhigh, standard tier) in Muse Code against Claude Code with Opus. Muse took 10–15 minutes versus Claude's 5–7 minutes, with both producing acceptable output. The slowdown came from more tool calls and longer pauses, not token speed.
A developer timed identical agent tasks building and deploying a static site to Cloudflare Workers, comparing Meta's muse-spark-1.3 (xhigh, standard tier) in Muse Code against Claude Code with Opus. Muse took 10–15 minutes versus Claude's 5–7 minutes, with both producing acceptable output. The slowdown came from more tool calls and longer pauses, not token speed.
1 engagement1 sourceredditSat, Sep 26, 2026, 10:50 PM
Releases·Hardware Release
Meta ditches the camera on its newest smart glasses
At Meta Connect 2026, Meta is launching Ray-Ban Meta Audio Glasses: smart glasses with Meta AI but no camera, positioned against growing public backlash over wearable surveillance tech. The timing lets users record nothing of others while still getting AI assistance, though Meta says the audio-only design was years in the making rather than a reaction.
At Meta Connect 2026, Meta is launching Ray-Ban Meta Audio Glasses: smart glasses with Meta AI but no camera, positioned against growing public backlash over wearable surveillance tech. The timing lets users record nothing of others while still getting AI assistance, though Meta says the audio-only design was years in the making rather than a reaction.
0 engagement1 sourcerssWed, Sep 23, 2026, 11:37 PM
Developments·Benchmark
Muse Wonder (high) scores 93.4 on the Extended NYT Connections Benchmark, ahead of Muse Spark 1.3 (high) at 85.1. It's among the most expensive models tested.
A Reddit post reports results from lechmazur's Extended NYT Connections Benchmark, where Muse Wonder (high) scored 93.4, ahead of Muse Spark 1.3 (high) at 85.1. Muse Wonder was also among the most expensive models tested on the benchmark.
A Reddit post reports results from lechmazur's Extended NYT Connections Benchmark, where Muse Wonder (high) scored 93.4, ahead of Muse Spark 1.3 (high) at 85.1. Muse Wonder was also among the most expensive models tested on the benchmark.
59 engagement1 sourceredditTue, Oct 13, 2026, 03:28 PM
Releases·Tool Releasefeatured
Meta launches Muse Code, a terminal coding agent for large code bases
Meta released Muse Code, a terminal-based AI coding agent currently in beta, aimed at helping programmers handle complex software engineering tasks across large repositories. The agent can plan changes, write code, and validate results, and is powered by Meta's AI models. It is available globally and can be installed with a single command.
Meta released Muse Code, a terminal-based AI coding agent currently in beta, aimed at helping programmers handle complex software engineering tasks across large repositories. The agent can plan changes, write code, and validate results, and is powered by Meta's AI models. It is available globally and can be installed with a single command.
Updated Mon, Aug 10, 2026, 07:58 AM — Binary analysis reveals hidden agent workflows and a Git plugin marketplace.
041317 engagement5 sourceshackernews, reddit, rssWed, Aug 5, 2026, 07:59 PM
Releases·Model Releasefeatured
Cat Jam (stealth model) is now free in OpenCode
Cat Jam, a stealth model, is now available for free in OpenCode. Posts describe it as fast, built for agentic coding, and offering zero data retention.
Cat Jam, a stealth model, is now available for free in OpenCode. Posts describe it as fast, built for agentic coding, and offering zero data retention.
2.4k engagement1 sourcexTue, Oct 13, 2026, 05:59 PM