Benchmark Almanac

MIVAS Bench

GPT-Live-1 + GPT-6 Astra 66.3% tops its leaderboard Grok Voice 3.0 56.5% leads the models we track
Not in model cards

Voice agents handle calls for a dermatology practice, a law firm, and a retailer. Each call must pass checks on tools, specialist handoffs, and database changes.

Bluejay ranks 14 model configurations on 216 held-out tasks. Its public dataset contains nine sample tasks.

Made by
Bluejay Labs
Released
August 2026
Kind
Customer service
Level
University
Scored
Checked automatically
Answers
Agent task
Tasks
Public, plus a private test set
Measures
Capability
Language
English
In and out
Audio, Text → Audio, Actions
Interaction
Multi-agent
Items from
Written by experts
Refresh
One fixed set
Run by
A host
Last model added
October 2026
Scores refreshed
November 20th, 2026
14
Entries on its own leaderboard, 4 models we track

GPT-Live-1 + GPT-6 Astra tops the leaderboard with 66.3% Grok Voice 3.0 leads with 56.5%

Best score per entry from the benchmark's own leaderboard, read on November 20th, 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 Grok Voice 3.0, at 56.5%. This view shows only models we track, each at its best.

  1. 1 GPT-Live-1 + GPT-6 Astra 66.3%
  2. 2 Phonic 1 (reasoning: high) 56.9%
  3. 3 Grok Voice 3.0 reasoning: high 56.5%
  4. 4 Gemini 3.8 Live Extended Thinking 55.5%
  5. 5 Grok Voice 2.0 53.7%
  6. 6 OpenAI Realtime 2.1 49.1%
  7. 7 Gemini 3.8 Live 43.5%
  8. 8 Amazon Nova Sonic 2.5 37.5%
  9. 9 Qwen Audio Realtime 36.1%
  10. 10 Gemini Flash Live 3.1 30.6%
  11. 11 Amazon Nova Sonic 2 27.8%
  12. 12 flux-general-en + GPT-4.1 + eleven_flash_v2_5 27.7%
  13. 13 Gemini 2.5 Flash Native Audio 18%
  14. 14 OpenAI Realtime 2.1 Mini 16.7%

Source: Official MIVAS Bench leaderboard

#EntryScoreSettingDateLinkSource
1 GPT-Live-1 + GPT-6 Astra 66.3% research.getbluejay.ai Official leaderboard
2 Phonic 1 (reasoning: high) 56.9% research.getbluejay.ai Official leaderboard
3 Grok Voice 3.0 56.5% reasoning: high Oct 2026 research.getbluejay.ai Official leaderboard
4 Gemini 3.8 Live Extended Thinking 55.5% Sep 2026 research.getbluejay.ai Official leaderboard
5 Grok Voice 2.0 53.7% Jul 2026 research.getbluejay.ai Official leaderboard
6 OpenAI Realtime 2.1 49.1% research.getbluejay.ai Official leaderboard
7 Gemini 3.8 Live 43.5% Sep 2026 research.getbluejay.ai Official leaderboard
8 Amazon Nova Sonic 2.5 37.5% research.getbluejay.ai Official leaderboard
9 Qwen Audio Realtime 36.1% research.getbluejay.ai Official leaderboard
10 Gemini Flash Live 3.1 30.6% research.getbluejay.ai Official leaderboard
11 Amazon Nova Sonic 2 27.8% research.getbluejay.ai Official leaderboard
12 flux-general-en + GPT-4.1 + eleven_flash_v2_5 27.7% research.getbluejay.ai Official leaderboard
13 Gemini 2.5 Flash Native Audio 18% research.getbluejay.ai Official leaderboard
14 OpenAI Realtime 2.1 Mini 16.7% research.getbluejay.ai Official leaderboard
#ModelScoreSettingDateLinkSource
1 Grok Voice 3.0 56.5% reasoning: high Oct 2026 research.getbluejay.ai Official leaderboard
2 Gemini 3.8 Live Extended Thinking 55.5% Sep 2026 research.getbluejay.ai Official leaderboard
3 Grok Voice 2.0 53.7% Jul 2026 research.getbluejay.ai Official leaderboard
4 Gemini 3.8 Live 43.5% Sep 2026 research.getbluejay.ai Official leaderboard

Rankings by part

We read 6 parts from the benchmark’s own leaderboard in 2 groups: industry and difficulty. Each ranks models on one part of the test. Gemini 3.8 Live Extended Thinking leads 2 of them, and 3 other models lead the rest. Arrows show how far a model moved from its overall rank among the models we track.

healthcare

4 models we track have a score here.

  1. 1 Gemini 3.8 Live Extended Thinking ↑1 63.9%
  2. 2 Grok Voice 3.0 ↓1 59.7%
  3. 3 Grok Voice 2.0 56.9%
  4. 4 Gemini 3.8 Live 48.6%

3 models have held number one on the official leaderboard since its release in August 2026

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

xAI Google DeepMind
Own leaderboard 3 leaders

Source: Official MIVAS Bench leaderboard

Scores over time

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

Plot with other benchmarks
On the benchmark’s own leaderboard Best so far
0% 20% 40% 60% 80% 100% Near the ceiling (85%) Grok Voice 3.0 Grok Voice 2.0 0% 20% 40% 60% 80% 100% Near the ceiling (85%) Grok Voice 3.0

Source: Official MIVAS Bench leaderboard

No lab model card reports it yet

We read the headline results of 540 launch posts from 57 labs, and none of them list this benchmark.

More in Tool use

All 265 in Tool use
  • AutomationBench In use Agent completes end-to-end business workflows inside a simulated company exposing dozens of REST APIs. Graded on resulting state. Gemini 4 Neon 56.3%
  • MCP Atlas In use Multi-step workflows over real MCP servers. Scores tool-use competency on realistic tasks. Gemini 4 Argon 90.7%
  • BFCL V4 In use Models call functions and tools, including multi-turn agentic tasks with web search and memory. Checked against expected calls and outcomes. Qwen3.7-Max 75%
  • τ³-bench Banking In use Agents act as banking customer-service reps over many turns with a simulated user, following policy. Pass^1 on the final database state. Qwen 3.8 Max 55.1%
  • Toolathlon-Verified In use 108 long-horizon tasks across 32 apps through about 600 MCP tools. Execution-based checkers inspect final environment state. GLM-5.3-Flash 78.4%
  • DeepSearchQA In use Questions that need multi-step search and synthesis. Final answers graded against references. Kimi K3 95%

Data audit

All 10 checks pass. Every benchmark

Catalog record 2 pass
  • Pass Task access Public tasks, plus a private test set.
  • Pass Confidence Checked against a primary source.
Scores 4 pass 1 note
  • Pass Independent runs Scores from the official leaderboard (14 entries).
  • Pass Models scored 4 models we track, plus 10 other entries on its leaderboard.
  • Pass Newest score The newest is from October 2026, for Grok Voice 3.0.
  • Note Frontier models None of its models are on the capabilities index.
  • Pass Headroom The best is 56.5%, under 85%, so it still separates models.
Leaderboard import 3 pass
  • Pass Importer We read the board on research.getbluejay.ai: 4 rows for models we track and 10 other entries at the last read.
  • Pass Last read November 20th, 2026.
  • Pass Read worked The last read worked.
Consistency 1 pass 1 note
  • Note Lab claims No model has both a lab number and an independent run.
  • Pass Known issues No published audit found problems.

How we check