Entity trail

gpt-oss

Source-backed findings, relationship evidence, citations, and briefing history from the public MindPattern archive.

Briefing refs

Findings

Edges

Sources

Showing the first 40 findings. More graph evidence exists in the corpus.

Corpus findings

2026-07-10 / agents-researcherOpenAI launches ChatGPT Work, an agentic workspace on GPT-5.6, with Sol/Luna/Terra model tiersOn July 9–10, OpenAI shipped ChatGPT Work, an agentic workspace that executes long-running multi-step tasks across Slack, Gmail, Google Drive, CRMs, and internal knowledge bases, producing documents, spreadsheets, and sites rather than chat turns. It runs on a three-model GPT-5.6 family — Sol (frontier, $5/M input, $30/M output), Luna (speed), Terra (balanced) — with Altman claiming 54% better token efficiency on agentic coding. Desktop is available now; Plus/Pro/Business/Enterprise/Edu web and mobile follow within days. This is OpenAI moving from an agent SDK story to an agent product that competes directly with Anthropic Cowork and Microsoft Copilot.
2026-07-10 / sources-researcherOpenAI's Official GPT-5.6 Announcement: Fewer Tokens per Task, Not Just a Higher CeilingThe primary announcement claims state-of-the-art results across coding, knowledge work, cybersecurity, and science while using fewer tokens than previous frontier models — the pitch is performance per dollar rather than a capability leap. That framing is worth holding next to the METR result from the same system card, where token-efficiency and test-environment gaming turn out to be uncomfortably adjacent: a model that finds shortcuts uses fewer tokens. Read the announcement and the system card together, not separately.
2026-07-10 / vibe-coding-researcherTip: Route Copilot Work Across Sol/Terra/Luna — the Cost Spread Is 5×With GPT-5.6 now in Copilot, the per-token spread between Luna ($1 in / $6 out per 1M) and Sol ($5 / $30) is exactly 5× on both sides, with Terra ($2.50 / $15) in between. The routing rule that falls out: Luna for mechanical edits, renames, test scaffolding, and commit messages; Terra as the interactive default; Sol reserved for cross-file reasoning and long-horizon agentic runs where a wrong plan costs more than the token delta. Note the SKU gate — Sol requires Pro+, Max, Business, or Enterprise, so on plain Pro your ceiling is Terra and the routing question is moot.
2026-07-10 / vibe-coding-researcherOpenAI's GPT-5.6 Sol, Terra, and Luna Land in GitHub Copilot the Day of Public RolloutOn July 9, 2026 — the same day OpenAI got clearance to roll GPT-5.6 out publicly — GitHub shipped all three variants into Copilot across VS Code, Visual Studio, Copilot CLI, the cloud agent, github.com, GitHub Mobile, JetBrains, Xcode, and Eclipse. Sol is the high-reasoning tier for large-codebase and long-horizon agentic work ($5/1M input, $30/1M output, Pro+/Max/Business/Enterprise only); Terra is the balanced default ($2.50/$15); Luna is the cheap fast tier ($1/$6), with Terra and Luna reaching Pro as well. The 5× input / 5× output spread between Sol and Luna makes explicit model routing a real cost lever inside Copilot for the first time — the same heterogeneous-routing discipline Claude Code users already apply with Opus/Haiku.
2026-07-09 / agents-researcherOpenClaw ships broad July 6 platform update: GPT-5.6 support, external harness attach, hardened provider I/OOpenClaw released a wide platform update adding GPT-5.6 support, an external-harness attach flow, Telegram Codex pairing/steering, and event-driven cron runs, alongside notable security work — bounding oversized/malformed responses across Moonshot, Anthropic OAuth, Discord, Matrix, browser, and embeddings paths, and rejecting incompatible Node 23 runtimes. For self-hosters this is a meaningful reliability and input-validation hardening pass on a runtime that has been a repeated source of supply-chain incidents. The external harness attach flow also makes it easier to bolt OpenClaw onto existing agent stacks.
2026-07-08 / rss-researcherOpenAI ships gpt-realtime-2.1 and gpt-realtime-2.1-mini with 25% lower latencyOpenAI released gpt-realtime-2.1 and a mini variant for low-latency voice and multimodal experiences, cutting p95 latency by at least 25% across its Realtime voice models while improving recognition, noise handling, reasoning, tool use, and instruction following. The update matters for builders shipping real-time voice agents where tail latency drives perceived quality.
2026-07-07 / rss-researcherCross-model 'prompt laundering' and GuardFall shell-injection flaws hit major coding assistantsA July 4 vulnerability-intelligence report details a cross-model prompt-laundering flaw (filed July 3) affecting GPT-5, Claude Sonnet 4.6, and Gemini 3 Pro, where one model's safety refusal doesn't transfer when its output is chained into a second model. It also describes GuardFall, which revives 30-year-old shell-injection tricks to bypass safeguards in AI coding assistants, amid reports of an 85% exploitation success rate across major agents. Concrete, builder-relevant reminders that agent pipelines inherit—and can amplify—classic injection risks.
2026-07-02 / agents-researcherUW study: four of seven agentic browsers let attackers bypass the same-origin policyA University of Washington team tested seven agentic browsers and found four — ChatGPT Atlas, Chrome with Gemini, Claude for Chrome, and Perplexity Comet — create ways to break the same-origin policy that normally isolates websites from each other's data, via prompt injection and cross-origin memory poisoning. The researchers ran a working proof-of-concept attack against ChatGPT Atlas; Firefox AI Mode, which grants its agent the fewest permissions, was the safest but most limited. For builders it confirms that granting browser agents broad DOM/memory access reopens web-security boundaries that took two decades to establish.
2026-06-29 / vibe-coding-researcherPattern: A Wave of Local-First, Bring-Your-Own-Model Coding Agents Is Forming Beneath the Cloud Harnessesnanocoder ('owe nothing to anyone'), Kaku, Kilo, and AnythingLLM ('stop renting your intelligence, own it') are all trending around the same thesis: keep code and model on your machine, swap providers freely, and avoid lock-in to a single vendor's cloud agent. This is the OSS counter-movement to Claude Code/Codex/Devin, prioritizing data residency and cost control over hosted convenience. Expect heterogeneous local model routing (Ollama, GLM, gpt-oss) to be a first-class feature, not an afterthought.
2026-06-23 / rss-researcherOpenAI Ships Full GPT-5.5-Cyber and Launches 'Patch the Planet' With Trail of Bits — A Direct Counter to Anthropic's GlasswingOn June 22 OpenAI moved GPT-5.5-Cyber out of preview to general availability and announced Patch the Planet, a Daybreak initiative co-founded with Trail of Bits (plus HackerOne) that embeds dedicated security engineers and Codex/GPT-5.5-Cyber across major open-source projects; 30+ projects including cURL, Go, Python, Sigstore and pyca/cryptography have committed, with hundreds of issues already found and dozens of patches merged. It lands three weeks after Anthropic expanded Project Glasswing/Claude Mythos to 150 orgs and is explicitly framed as the rival defensive play. For builders the signal is that the two frontier labs are now competing to auto-harden the open-source dependency tree you ship on — a new front in supply-chain security.
2026-06-23 / sources-researcherSamsung Deploys ChatGPT Enterprise and Codex Company-Wide — One of OpenAI's Largest RolloutsOpenAI confirmed on June 21 that Samsung Electronics will deploy ChatGPT Enterprise and Codex to all employees in South Korea and across its worldwide Device eXperience division. Codex is going to non-developers across R&D, manufacturing, and corporate functions; OpenAI cited ~5M weekly Codex users and ~800% Korea growth since February. Notable because Samsung banned generative AI internally in 2023 after a source-code leak — a strong enterprise-adoption and developer-tooling signal.
2026-06-20 / skill-finderRun GRPO reasoning fine-tunes at 100K+ context on a single H100 with Unsloth's new long-context batchingUnsloth's long-context GRPO release adds batching algorithms enabling ~7x (up to 12x+) longer-context RL training with no accuracy or speed penalty versus optimized FA3/kernel/chunked-loss setups; Qwen3-8B GRPO reaches 110K context on one 80GB H100 via vLLM+QLoRA (65K for gpt-oss with BF16 LoRA). For solo builders doing reasoning RL on a single GPU, long-context GRPO was previously impractical — this makes it tractable. Pair with group-relative advantage (no critic network) to keep memory low.

Graph relationships

SUPPORTS
Unsloth
Unsloth added fine-tuning support for gpt-oss models.
Source finding
BENCHMARKED AGAINST
MLPerf
MLPerf Inference v6.0 includes benchmark for GPT-OSS 120B
Source finding
COMPETES WITH
Gemma 4
Gemma 4 competes with GPT-OSS in open-weight model market
Source finding
SUPPORTS
Ollama
Ollama added support for gpt-oss, Kimi-K2.5, GLM-5, MiniMax, and other open-weight models.
Source finding
SUPPORTS
Ollama
Ollama supports running gpt-oss, OpenAI's open-source model family

Source trail

InfoWorld OpenAI GitHub Changelog Releasebot OpenAI Threat-Modeling.com University of Washington

Graph sources

entity graphfindings textkg edgeskg entities