Timeline · 2024-2026

AI coding tool releases

Every major AI coding tool launch, feature, model release, and ecosystem milestone, dated, sourced, in reverse chronological order. Curated and updated as new releases ship.

LaunchFeatureModelStandardEcosystem

202610 releases

  1. LaunchGoogle

    Google Antigravity 2.0

    Multi-agent desktop platform launches at Google I/O, desktop app + Go CLI + open-source SDK, powered by Gemini 3.5 Flash.

    Why it matters: First multi-agent-native platform from a frontier provider. Native agent teams in the architecture rather than bolted on.

  2. ModelGoogle

    Gemini 3.5 Flash

    Google's fast/cheap variant of Gemini 3 family launches alongside Antigravity 2.0.

    Why it matters: Sub-300ms first token; the speed/cost variant for inline completion workloads.

  3. FeatureAnysphere

    Cursor Composer 2.5

    Cursor ships Composer 2.5 with improved model switching and longer agent runs.

    Why it matters: Mid-cycle refinement of Cursor's agent mode, narrows the gap with Claude Code on long autonomous runs.

  4. ModelAnthropic

    Claude Opus 4.7 (1M context)

    Anthropic's flagship coding model with 1M-token context becomes generally available.

    Why it matters: 1M context for multi-file refactors and big-repo work. Tops SWE-bench's agentic + multi-file variants.

  5. ModelOpenAI

    GPT-5.5

    OpenAI ships GPT-5.5, incremental on GPT-5 with better tool use and reduced hallucination.

    Why it matters: Stays competitive on SWE-bench single-file + speed variants; default in Codex CLI.

  6. ModelGoogle

    Gemini 3.1 Pro

    Google's flagship coding model, 1M+ context default, 2M+ in some configurations.

    Why it matters: Longest context window of the three April flagships. Strong on long-context reasoning and multimodal.

  7. FeatureAnthropic

    Claude Code /loop

    Scheduled-task feature ships, cron-style background runs of skills, no terminal required.

    Why it matters: Turned Claude Code from a chat tool into a background worker. Nightly audits, daily PRs, pre-standup pipelines now built in.

  8. FeatureAnthropic

    Claude Code /voice

    Push-to-talk voice mode in 20 languages with streaming transcription.

    Why it matters: Small feature, rhythm-changing for narration and PR descriptions. Sleeper hit of the March release.

  9. FeatureAnthropic

    Claude Code Sub-agents (GA)

    Sub-agents (parallel isolated agents with their own context windows) reach general availability.

    Why it matters: The headline 2026 shift. Multi-agent orchestration becomes a first-class pattern in Claude Code.

  10. LaunchAWS

    AWS Kiro launch

    Amazon launches Kiro, the spec-driven AI IDE powered by Bedrock.

    Why it matters: Spec-driven development as a first-class IDE pattern. AWS's answer to Cursor + Claude Code.

20256 releases

  1. EcosystemAnthropic ecosystem

    MCP registry passes 5,000 servers

    Public MCP server registry crosses 5,000 entries, 4x the early-2025 count.

    Why it matters: MCP wins the connectivity layer. ChatGPT Apps SDK targets a different surface (consumer ChatGPT distribution).

  2. Ecosystemskills-hub.ai

    Skills-hub.ai catalog hits 4,000 skills

    The open registry's catalog crosses 4,000 skills synced from 80+ official sources.

    Why it matters: Critical mass for the SKILL.md standard. Cross-tool portability becomes a real choice point.

  3. LaunchCline Bot Inc.

    Cline 1.0

    Open-source VS Code agent reaches 1.0, per-step approval UI matures and MCP support lands.

    Why it matters: First open-source agent at production polish. The 'see every action before it runs' UX shipped.

  4. FeatureAnysphere

    Cursor Composer + agent mode

    Cursor's agent mode reaches feature parity with Composer for multi-file work.

    Why it matters: Cursor closes the gap with Claude Code on multi-file refactors while keeping its inline-completion lead.

  5. LaunchCodeium

    Windsurf 1.0

    Codeium's AI-first IDE ships 1.0 with Cascade agent flows.

    Why it matters: Cascade introduces a flow-first agent UX, distinct from Composer-style chat agent. Price/feature sweet spot at $15/mo.

  6. LaunchOpenAI

    OpenAI Codex CLI (Apache 2.0)

    OpenAI open-sources Codex CLI under Apache 2.0.

    Why it matters: First major frontier provider to ship an Apache 2.0 terminal agent. Sets template Gemini CLI follows later.

20242 releases

  1. StandardAnthropic

    Model Context Protocol (MCP) 1.0

    Anthropic publishes MCP 1.0, open standard for AI clients to call external tools.

    Why it matters: Becomes the lingua franca of AI coding tools through 2025-2026. 30+ clients adopt it natively.

  2. LaunchAnthropic

    Claude Code launch

    Anthropic launches Claude Code, terminal-native agentic coding assistant.

    Why it matters: First major terminal-native coding agent from a frontier provider. Sets template for Codex CLI, Gemini CLI, Antigravity CLI.

Stay current as the field moves

New releases land in this timeline as they ship. Get them weekly via the newsletter.

Subscribe →Blog