Self-hosted, Self-Improving, Federated Everything App — dev management, agent orchestration, persistent mind, digital twin, second brain, and creative pipeline from a single dashboard over Tailscale to port the user into the OS
JavaScript ai-agentsdashboarddeveloper-toolsdigital-twinquantified-selfsecond-brainself-hostedtailscale
feat: add Artificial Analysis v4.3.2 scores for Claude Haiku 5.5 to model comparison
Adam Eivy136b5a39a3ed
diff --git a/data.reference/model-comparison.json b/data.reference/model-comparison.jsonindex 355b5761df..40c9c8587b 100644--- a/data.reference/model-comparison.json+++ b/data.reference/model-comparison.json@@ -58253,6 +58253,271 @@}},"notes": "Publisher result for the base checkpoint; local quantization and hardware are unmeasured. Keep this harness-specific score separate from the AA intelligence index."+ },+ {+ "id": "aa-v4.3.2-anthropic-claude-haiku-5.5-low",+ "provider": "Anthropic",+ "model": "claude-haiku-5.5",+ "effort": "low",+ "configuration": "Published API model; Reasoning, Low Effort, Default Fallback",+ "billing": "api",+ "benchmark": "Artificial Analysis Intelligence Index v4.3.2",+ "quality": {+ "value": 29,+ "source": {+ "url": "https://artificialanalysis.ai/models/claude-haiku-5-5-low",+ "retrievedAt": "2026-10-10T00:00:00.000Z",+ "methodology": "Artificial Analysis Intelligence Index v4.3.2 composite score (published as a whole number)."+ }+ },+ "costPerTask": {+ "value": 0.02,+ "source": {+ "url": "https://artificialanalysis.ai/models/claude-haiku-5-5-low",+ "retrievedAt": "2026-10-10T00:00:00.000Z",+ "methodology": "Artificial Analysis Intelligence Index v4.3.2; cost per task evaluation (published rounded to the cent)."+ }+ },+ "inputPerMillion": {+ "value": 0.1,+ "source": {+ "url": "https://artificialanalysis.ai/models/claude-haiku-5-5-low",+ "retrievedAt": "2026-10-10T00:00:00.000Z",+ "methodology": "Published USD per million uncached input tokens."+ }+ },+ "outputPerMillion": {+ "value": 0.5,+ "source": {+ "url": "https://artificialanalysis.ai/models/claude-haiku-5-5-low",+ "retrievedAt": "2026-10-10T00:00:00.000Z",+ "methodology": "Published USD per million output tokens."+ }+ },+ "reasoningPerMillion": null,+ "responseSeconds": null,+ "tokensPerSecond": {+ "value": 181.4,+ "source": {+ "url": "https://artificialanalysis.ai/models/claude-haiku-5-5-low",+ "retrievedAt": "2026-10-10T00:00:00.000Z",+ "methodology": "Artificial Analysis output tokens per second on the first-party Anthropic API."+ }+ },+ "quota": null,+ "notes": "Sourced from Artificial Analysis (Claude Haiku 5.5 (Low)). Independent measurement; supersedes the publisher-launch-eval calibration for this effort. End-to-end response time is not published for this model."+ },+ {+ "id": "aa-v4.3.2-anthropic-claude-haiku-5.5-medium",+ "provider": "Anthropic",+ "model": "claude-haiku-5.5",+ "effort": "medium",+ "configuration": "Published API model; Reasoning, Medium Effort, Default Fallback",+ "billing": "api",+ "benchmark": "Artificial Analysis Intelligence Index v4.3.2",+ "quality": {+ "value": 34,+ "source": {+ "url": "https://artificialanalysis.ai/models/claude-haiku-5-5-medium",+ "retrievedAt": "2026-10-10T00:00:00.000Z",+ "methodology": "Artificial Analysis Intelligence Index v4.3.2 composite score (published as a whole number)."+ }+ },+ "costPerTask": {+ "value": 0.05,+ "source": {+ "url": "https://artificialanalysis.ai/models/claude-haiku-5-5-medium",+ "retrievedAt": "2026-10-10T00:00:00.000Z",+ "methodology": "Artificial Analysis Intelligence Index v4.3.2; cost per task evaluation (published rounded to the cent)."+ }+ },+ "inputPerMillion": {+ "value": 0.1,+ "source": {+ "url": "https://artificialanalysis.ai/models/claude-haiku-5-5-medium",+ "retrievedAt": "2026-10-10T00:00:00.000Z",+ "methodology": "Published USD per million uncached input tokens."+ }+ },+ "outputPerMillion": {+ "value": 0.5,+ "source": {+ "url": "https://artificialanalysis.ai/models/claude-haiku-5-5-medium",+ "retrievedAt": "2026-10-10T00:00:00.000Z",+ "methodology": "Published USD per million output tokens."+ }+ },+ "reasoningPerMillion": null,+ "responseSeconds": null,+ "tokensPerSecond": {+ "value": 170.4,+ "source": {+ "url": "https://artificialanalysis.ai/models/claude-haiku-5-5-medium",+ "retrievedAt": "2026-10-10T00:00:00.000Z",+ "methodology": "Artificial Analysis output tokens per second on the first-party Anthropic API."+ }+ },+ "quota": null,+ "notes": "Sourced from Artificial Analysis (Claude Haiku 5.5 (Medium)). Independent measurement; supersedes the publisher-launch-eval calibration for this effort. End-to-end response time is not published for this model."+ },+ {+ "id": "aa-v4.3.2-anthropic-claude-haiku-5.5-high",+ "provider": "Anthropic",+ "model": "claude-haiku-5.5",+ "effort": "high",+ "configuration": "Published API model; Reasoning, High Effort, Default Fallback",+ "billing": "api",+ "benchmark": "Artificial Analysis Intelligence Index v4.3.2",+ "quality": {+ "value": 38,+ "source": {+ "url": "https://artificialanalysis.ai/models/claude-haiku-5-5-high",+ "retrievedAt": "2026-10-10T00:00:00.000Z",+ "methodology": "Artificial Analysis Intelligence Index v4.3.2 composite score (published as a whole number)."+ }+ },+ "costPerTask": {+ "value": 0.08,+ "source": {+ "url": "https://artificialanalysis.ai/models/claude-haiku-5-5-high",+ "retrievedAt": "2026-10-10T00:00:00.000Z",+ "methodology": "Artificial Analysis Intelligence Index v4.3.2; cost per task evaluation (published rounded to the cent)."+ }+ },+ "inputPerMillion": {+ "value": 0.1,+ "source": {+ "url": "https://artificialanalysis.ai/models/claude-haiku-5-5-high",+ "retrievedAt": "2026-10-10T00:00:00.000Z",+ "methodology": "Published USD per million uncached input tokens."+ }+ },+ "outputPerMillion": {+ "value": 0.5,+ "source": {+ "url": "https://artificialanalysis.ai/models/claude-haiku-5-5-high",+ "retrievedAt": "2026-10-10T00:00:00.000Z",+ "methodology": "Published USD per million output tokens."+ }+ },+ "reasoningPerMillion": null,+ "responseSeconds": null,+ "tokensPerSecond": {+ "value": 184.4,+ "source": {+ "url": "https://artificialanalysis.ai/models/claude-haiku-5-5-high",+ "retrievedAt": "2026-10-10T00:00:00.000Z",+ "methodology": "Artificial Analysis output tokens per second on the first-party Anthropic API."+ }+ },+ "quota": null,+ "notes": "Sourced from Artificial Analysis (Claude Haiku 5.5 (High)). Independent measurement; supersedes the publisher-launch-eval calibration for this effort. End-to-end response time is not published for this model."+ },+ {+ "id": "aa-v4.3.2-anthropic-claude-haiku-5.5-xhigh",+ "provider": "Anthropic",+ "model": "claude-haiku-5.5",+ "effort": "xhigh",+ "configuration": "Published API model; Reasoning, Xhigh Effort, Default Fallback",+ "billing": "api",+ "benchmark": "Artificial Analysis Intelligence Index v4.3.2",+ "quality": {+ "value": 41,+ "source": {+ "url": "https://artificialanalysis.ai/models/claude-haiku-5-5-xhigh",+ "retrievedAt": "2026-10-10T00:00:00.000Z",+ "methodology": "Artificial Analysis Intelligence Index v4.3.2 composite score (published as a whole number)."+ }+ },+ "costPerTask": {+ "value": 0.12,+ "source": {+ "url": "https://artificialanalysis.ai/models/claude-haiku-5-5-xhigh",+ "retrievedAt": "2026-10-10T00:00:00.000Z",+ "methodology": "Artificial Analysis Intelligence Index v4.3.2; cost per task evaluation (published rounded to the cent)."+ }+ },+ "inputPerMillion": {+ "value": 0.1,+ "source": {+ "url": "https://artificialanalysis.ai/models/claude-haiku-5-5-xhigh",+ "retrievedAt": "2026-10-10T00:00:00.000Z",+ "methodology": "Published USD per million uncached input tokens."+ }+ },+ "outputPerMillion": {+ "value": 0.5,+ "source": {+ "url": "https://artificialanalysis.ai/models/claude-haiku-5-5-xhigh",+ "retrievedAt": "2026-10-10T00:00:00.000Z",+ "methodology": "Published USD per million output tokens."+ }+ },+ "reasoningPerMillion": null,+ "responseSeconds": null,+ "tokensPerSecond": {+ "value": 188.3,+ "source": {+ "url": "https://artificialanalysis.ai/models/claude-haiku-5-5-xhigh",+ "retrievedAt": "2026-10-10T00:00:00.000Z",+ "methodology": "Artificial Analysis output tokens per second on the first-party Anthropic API."+ }+ },+ "quota": null,+ "notes": "Sourced from Artificial Analysis (Claude Haiku 5.5 (Xhigh)). Independent measurement; supersedes the publisher-launch-eval calibration for this effort. End-to-end response time is not published for this model."+ },+ {+ "id": "aa-v4.3.2-anthropic-claude-haiku-5.5-max",+ "provider": "Anthropic",+ "model": "claude-haiku-5.5",+ "effort": "max",+ "configuration": "Published API model; Reasoning, Max Effort, Default Fallback",+ "billing": "api",+ "benchmark": "Artificial Analysis Intelligence Index v4.3.2",+ "quality": {+ "value": 43,+ "source": {+ "url": "https://artificialanalysis.ai/models/claude-haiku-5-5",+ "retrievedAt": "2026-10-10T00:00:00.000Z",+ "methodology": "Artificial Analysis Intelligence Index v4.3.2 composite score (published as a whole number)."+ }+ },+ "costPerTask": {+ "value": 0.21,+ "source": {+ "url": "https://artificialanalysis.ai/models/claude-haiku-5-5",+ "retrievedAt": "2026-10-10T00:00:00.000Z",+ "methodology": "Artificial Analysis Intelligence Index v4.3.2; cost per task evaluation (published rounded to the cent)."+ }+ },+ "inputPerMillion": {+ "value": 0.1,+ "source": {+ "url": "https://artificialanalysis.ai/models/claude-haiku-5-5",+ "retrievedAt": "2026-10-10T00:00:00.000Z",+ "methodology": "Published USD per million uncached input tokens."+ }+ },+ "outputPerMillion": {+ "value": 0.5,+ "source": {+ "url": "https://artificialanalysis.ai/models/claude-haiku-5-5",+ "retrievedAt": "2026-10-10T00:00:00.000Z",+ "methodology": "Published USD per million output tokens."+ }+ },+ "reasoningPerMillion": null,+ "responseSeconds": null,+ "tokensPerSecond": {+ "value": 240.1,+ "source": {+ "url": "https://artificialanalysis.ai/models/claude-haiku-5-5",+ "retrievedAt": "2026-10-10T00:00:00.000Z",+ "methodology": "Artificial Analysis output tokens per second on the first-party Anthropic API."+ }+ },+ "quota": null,+ "notes": "Sourced from Artificial Analysis (Claude Haiku 5.5 (Max)). Independent measurement; supersedes the publisher-launch-eval calibration for this effort. End-to-end response time is not published for this model."}]}