Archive/State of the Index — October 2026

Monthly Report

State of the Index — October 2026

Published November 1, 2026·RSS

The Ghost Index opened October at 52.9%, continuing directly from September's close, then dipped to its monthly low of 52.6% on October 16 before recovering steadily through the second half. It closed October 31 at 53.8% — a net gain of 0.9 percentage points, reversing September's −1.2 point decline and recording the index's first positive month since July. The indexed base grew by 10 agents (1,447 → 1,457) and alive agents expanded by 18 (766 → 784), indicating that October's recovery reflected genuine activity gains rather than base contraction. Service held at 77.6% (66 of 85); Developer gained to 52.0% (693 of 1,332), its highest reading since August.

October's headline board move was on the Developer ranking: openclaw reversed its September run, falling from #1 to #3 (74.5→70.96) as OpenAI Agents Python claimed the top slot (73.82) and CrewAI reclaimed #2 (71.81). The five-slot spread compressed to 6.7 points — the tightest the board has been, reflecting narrowing distance between the leading frameworks. On Model Families, DeepSeek and Mistral traded places: DeepSeek held its score (72) and rose to #3 while Mistral dropped 2 points (73→71) to #4; Qwen slipped 2 points but retained the top (80). On Tokenized, October saw broad compression — AIXBT lost 1 point (82), Ribbita lost 2 points (73), Luna lost 1 point (64), and Voice of the Gods dropped out; 717ai entered at #5 with 59. The Service top three held for the second consecutive month (A2A 77, a2a-python 74, evolver 73); agent-teams-ai re-entered at #5.

The evidence-ranked tier reached 242 agents at October close — 30 more than September's 212, the largest single-month expansion since August's +29. The archive compounded to 223,497 daily snapshots as of November 1, adding approximately 45,020 new records in October alone. Every snapshot is Merkle-anchored to Base at 04:30 UTC the following morning and independently recomputable via /oracle. October marks month seven of an unbroken nightly snapshot run since April 2026.

All four boards at October close

Live from /api/rankings/*/llm-summary as of November 1, 2026.

Developer

GitHub · package usage · ecosystem signal
v2.c-public · full ranking →
1OpenAI Agents Python73.82
2CrewAI71.81
3openclaw70.96
4a2a-python67.35
5Google ADK Python67.15

The month's headline move: openclaw reversed its September run — fell from #1 (74.5) to #3 (70.96) as OpenAI Agents Python rose from #2 to #1 (73.82) and CrewAI reclaimed #2 (71.81). a2a-python rose from #5 to #4; Google ADK Python fell from #4 to #5. The five-slot spread compressed to 6.7 points — the tightest the Developer board has been, signaling narrowing distance between leading frameworks.

Model Families

HuggingFace · LMArena · deployment breadth
v1.4-with-deployment · full ranking →
1Alibaba Qwen80
2Google Gemini79
3DeepSeek72
4Mistral71
5Meta Llama70

DeepSeek and Mistral traded places: DeepSeek held its score (72) and rose from #4 to #3 as Mistral dropped 2 points (73→71) and fell to #4. Qwen slipped 2 points (82→80) but retained the top position. Gemini held at 79 (#2) and Meta Llama was unchanged at 70 (#5). The order is otherwise the same as September with two families swapped in the middle.

Tokenized

market cap · liquidity · TVL
v1.1-tokenized-tvl · full ranking →
1AIXBT82
2Ribbita73
3G.A.M.E66
4Luna64
5717ai59

October saw broad compression across the board: AIXBT lost 1 point (83→82), Ribbita lost 2 points (75→73), Luna lost 1 point (65→64), G.A.M.E held at 66. Voice of the Gods (ADM, 61) dropped out; 717ai (WIRE) entered at #5 with 59. AIXBT holds #1 on social signal strength despite Ribbita's larger raw market cap. The liveness column still reads 0% — a standing instrument gap: the probe listens on HTTP while these agents live on-chain.

Service

adoption · source quality · activity
v1.1-service-forks · full ranking →
1A2A77
2a2a-python74
3evolver73
4bitterbot-desktop70
5agent-teams-ai70

The top three held for the second consecutive month (A2A 77, a2a-python 74, evolver 73). a2a-samples, which returned to #4 in September, dropped out again; bitterbot-desktop rose from #5 to #4 (score unchanged at 70). agent-teams-ai re-entered at #5 — the same slot it held in August. The A2A protocol cluster still occupies the top four positions.

Ghost Index: 53.8% — up 0.9 points from September

53.8%

784 alive · 673 ghosts · 1,457 indexed · Oct 1 open: 52.9%

+0.9 points across October. Dipped to 52.6% on Oct 16, then recovered to hold a 53.4–54.0% range through the second half. First positive month since July. Every number here is independently recomputable via /oracle.

MCP servers*15 / 15100%
Model families10 / 10100%
Service66 / 8577.6%
Developer693 / 133252.0%
Tokenized*0 / 150%

* Two standing flags. MCP's 100% reflects a selection effect — we track actively maintained servers. Tokenized's 0% is an instrument gap: the probe listens on HTTP while those agents live on-chain. The fix is in progress; when the number moves it will move because the measurement improved, stated in those words.

October 2026 in data

1,457

Agents indexed

53.8%

Ghost Index liveness

223,497

Snapshots archived

242

Evidence-ranked (Oct close)

The index runs autonomously — daily snapshots at 02:00 UTC, nightly liveness scoring at 23:50 UTC, Sunday ranking runs, and Merkle anchoring to Base at 04:30 UTC. Every number here is live and independently recomputable via /oracle. The archive is the product: each nightly snapshot is a timestamped record of the agent economy that cannot be backfilled. The data is available machine-readable at /api/ghost-index/v1, /api/rankings/*/llm-summary, and /api/agent-economy/llm-summary.

All Rankings →Ghost Index →Daily Changes →Methodology →RSS →