Benchmark Almanac

Echo General AI Prediction Leaderboard

EchoZ-1.0 1,026 tops its leaderboard DeepSeek-V4-Pro 1,018 leads the models we track Not in model cards

Models forecast unresolved events drawn from prediction markets, synthesized news questions and expert-written questions. Forecasts made at aligned times are compared in pairwise battles, and an Elo fit ranks the models.

Domains include finance, politics, crypto, sports and esports. In the 2026-03-27 snapshot UniPat's own EchoZ-1.0 led 11 models, ahead of Gemini 3.1 Pro.

Made by UniPat AI and Peking University Released Mar 2026 Real-world events Tasks kept private Last model added Apr 2026 Scores refreshed 2 Nov 2026
Level
Expert Specialists with years of training, often over hours of work
Scored
Settled by events Predictions are scored against what later happens.
Answers
Prediction Probabilities for events that have not happened yet.
Leaderboard unipat.ai
18
Entries on its own leaderboard, 17 models we track

EchoZ-1.0 tops the leaderboard with 1,026

Best score per entry from the benchmark's own leaderboard, read on 2 Nov 2026. Where a model ran with several settings or harnesses, we keep its best. Names in gray aren't in our model list: fine-tunes, agent systems, products and older models. They don't count toward leaders. The best model we track is DeepSeek-V4-Pro, at 1,018.

The newest model with a score here came out in Apr 2026, and nobody has tested a newer one. Among the 20 highest-ranked models on the capability index, these have no score yet:

18 entries. Scroll sideways for the rest. Bars start at 900.

  1. 1,026 EchoZ-1.0
  2. 1,018 DeepSeek-V4-Pro
  3. 1,017 Kimi K2.6
  4. 1,009 Claude Opus 4.6
  5. 1,006 Gemini 3.1 Pro
  6. 1,006 GLM-5.1
  7. 1,003 Grok 4.1 Fast
  8. 1,002 MiniMax-M2.7
  9. 1,001 GPT-5.2
  10. 1,001 Grok 4.20
  11. 999 GPT-5.4
  12. 996 Seed2.0 Pro
  13. 986 Qwen3.5 397B-A17B
  14. 984 GLM-5
  15. 980 MiniMax-M2.5
  16. 967 Qwen3-Max
  17. 964 DeepSeek-V3.2
  18. 920 Kimi K2.5

Bars start at 900.

  1. 1 EchoZ-1.0 1,026
  2. 2 DeepSeek-V4-Pro 1,018
  3. 3 Kimi K2.6 1,017
  4. 4 Claude Opus 4.6 1,009
  5. 5 Gemini 3.1 Pro 1,006
  6. 6 GLM-5.1 1,006
  7. 7 Grok 4.1 Fast 1,003
  8. 8 MiniMax-M2.7 1,002
  9. 9 GPT-5.2 1,001
  10. 10 Grok 4.20 1,001
  11. 11 GPT-5.4 999
  12. 12 Seed2.0 Pro 996
  13. 13 Qwen3.5 397B-A17B 986
  14. 14 GLM-5 984
  15. 15 MiniMax-M2.5 980
  16. 16 Qwen3-Max 967
  17. 17 DeepSeek-V3.2 964
  18. 18 Kimi K2.5 920
#EntryScoreSettingDateSource
1 EchoZ-1.0 1,026 Official leaderboard
2 DeepSeek-V4-Pro 1,018 Apr 2026 Official leaderboard
3 Kimi K2.6 1,017 Apr 2026 Official leaderboard
4 Claude Opus 4.6 1,009 Feb 2026 Official leaderboard
5 Gemini 3.1 Pro 1,006 Feb 2026 Official leaderboard
6 GLM-5.1 1,006 Apr 2026 Official leaderboard
7 Grok 4.1 Fast 1,003 Nov 2025 Official leaderboard
8 MiniMax-M2.7 1,002 Mar 2026 Official leaderboard
9 GPT-5.2 1,001 Dec 2025 Official leaderboard
10 Grok 4.20 1,001 Feb 2026 Official leaderboard
11 GPT-5.4 999 Mar 2026 Official leaderboard
12 Seed2.0 Pro 996 Feb 2026 Official leaderboard
13 Qwen3.5 397B-A17B 986 Feb 2026 Official leaderboard
14 GLM-5 984 Feb 2026 Official leaderboard
15 MiniMax-M2.5 980 Feb 2026 Official leaderboard
16 Qwen3-Max 967 Sep 2025 Official leaderboard
17 DeepSeek-V3.2 964 Dec 2025 Official leaderboard
18 Kimi K2.5 920 Jan 2026 Official leaderboard

3 models have held number one since its release in Mar 2026

Lab claims count from the day the card came out, independent runs from the model’s release.

Anthropic Moonshot AI DeepSeek
Own leaderboard Not tested since Apr 2026
DeepSeek-V4-Pro 1,018

Scores over time

All the scores we have, one dot each, so a model tested at several settings appears more than once.

On the benchmark’s own leaderboard
920 940 960 980 1,000 1,020 2026 920 940 960 980 1,000 1,020 2026

No lab model card reports it yet

We read the headline results of 479 launch posts from 51 labs, and none of them list this benchmark.

More in Forecasting

All 12 in Forecasting
  • ForecastBench In use Models give probabilities for 500 auto-generated questions per round about future events, drawn from prediction markets and time-series datasets. Forecasts are scored with difficulty-adjusted Brier scores once the questions resolve. Claude 3 Haiku 44.7
  • Prophet Arena In use Models and agents forecast open Kalshi prediction-market events and are scored once the events resolve. The live board ranks a difficulty-adjusted Arena score and also reports calibration error and market return. Claude Opus 5.5 86.0
  • Human Behavior Bench Not in model cards Models predict how specified groups of U.S. adults answer closed survey questions, as a distribution over options. Scored as 1 minus total variation distance. Gemini 4 Argon 80.6
  • BTF-3 Not in model cards Agents forecast already-resolved binary and numeric questions while researching a frozen snapshot of the web. Scoring pools Brier and ranked probability scores. Claude Opus 5 0.12
  • BackSearch Not in model cards Paid API that searches a frozen web archive as of a given date, for forecasting backtests and reproducible web-enabled evals. A tool, not a benchmark.
  • BallotBench Not in model cards Nine frontier agents forecast 863 contests in the 2026 US midterms each day until polls close. Proper scoring rules grade forecasts against certified results, beside Kalshi, Polymarket and expert ratings.