Skip to content

Commit 325b5cc

Browse files
nicoloboschinicoloboschi
authored
feat: switch default model to gpt-4o-mini (#410)
1 parent 0758827 commit 325b5cc

File tree

28 files changed

+89
-112
lines changed

28 files changed

+89
-112
lines changed

‎.env.example‎

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -5,7 +5,7 @@
55
# Supported providers: openai, groq, ollama, gemini, anthropic, lmstudio, vertexai
66
HINDSIGHT_API_LLM_PROVIDER=openai
77
HINDSIGHT_API_LLM_API_KEY=your-api-key-here
8-
HINDSIGHT_API_LLM_MODEL=o3-mini
8+
HINDSIGHT_API_LLM_MODEL=gpt-4o-mini
99
HINDSIGHT_API_LLM_BASE_URL=https://api.openai.com/v1
1010

1111
# Example: Anthropic Claude configuration

‎.github/workflows/test.yml‎

Lines changed: 42 additions & 61 deletions
Original file line numberDiff line numberDiff line change
@@ -171,10 +171,9 @@ jobs:
171171
test-rust-cli:
172172
runs-on: ubuntu-latest
173173
env:
174-
HINDSIGHT_API_LLM_PROVIDER: groq
175-
HINDSIGHT_API_LLM_API_KEY: ${{ secrets.GROQ_API_KEY }}
176-
HINDSIGHT_API_LLM_MODEL: openai/gpt-oss-20b
177-
HINDSIGHT_API_LLM_GROQ_SERVICE_TIER: flex
174+
HINDSIGHT_API_LLM_PROVIDER: openai
175+
HINDSIGHT_API_LLM_API_KEY: ${{ secrets.OPENAI_API_KEY }}
176+
HINDSIGHT_API_LLM_MODEL: gpt-4o-mini
178177
HINDSIGHT_API_URL: http://localhost:8888
179178
GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
180179
UV_INDEX: pytorch=https://download.pytorch.org/whl/cpu
@@ -234,7 +233,6 @@ jobs:
234233
HINDSIGHT_API_LLM_PROVIDER=${{ env.HINDSIGHT_API_LLM_PROVIDER }}
235234
HINDSIGHT_API_LLM_API_KEY=${{ env.HINDSIGHT_API_LLM_API_KEY }}
236235
HINDSIGHT_API_LLM_MODEL=${{ env.HINDSIGHT_API_LLM_MODEL }}
237-
HINDSIGHT_API_LLM_GROQ_SERVICE_TIER=${{ env.HINDSIGHT_API_LLM_GROQ_SERVICE_TIER }}
238236
EOF
239237
240238
- name: Start API server
@@ -345,7 +343,9 @@ jobs:
345343
- name: Smoke test - verify container starts
346344
if: matrix.variant == 'slim'
347345
env:
348-
GROQ_API_KEY: ${{ secrets.GROQ_API_KEY }}
346+
HINDSIGHT_API_LLM_PROVIDER: openai
347+
HINDSIGHT_API_LLM_API_KEY: ${{ secrets.OPENAI_API_KEY }}
348+
HINDSIGHT_API_LLM_MODEL: gpt-4o-mini
349349
HINDSIGHT_API_EMBEDDINGS_PROVIDER: openai
350350
HINDSIGHT_API_EMBEDDINGS_OPENAI_API_KEY: ${{ secrets.OPENAI_API_KEY }}
351351
HINDSIGHT_API_RERANKER_PROVIDER: cohere
@@ -355,15 +355,13 @@ jobs:
355355
test-api:
356356
runs-on: ubuntu-latest
357357
env:
358-
HINDSIGHT_API_LLM_PROVIDER: groq
359-
HINDSIGHT_API_LLM_API_KEY: ${{ secrets.GROQ_API_KEY }}
360-
GROQ_API_KEY: ${{ secrets.GROQ_API_KEY }}
358+
HINDSIGHT_API_LLM_PROVIDER: openai
359+
HINDSIGHT_API_LLM_API_KEY: ${{ secrets.OPENAI_API_KEY }}
361360
GEMINI_API_KEY: ${{ secrets.GEMINI_API_KEY }}
362361
OPENAI_API_KEY: ${{ secrets.OPENAI_API_KEY }}
363362
COHERE_API_KEY: ${{ secrets.COHERE_API_KEY }}
364363
HINDSIGHT_API_EMBEDDINGS_OPENAI_API_KEY: ${{ secrets.OPENAI_API_KEY }}
365-
HINDSIGHT_API_LLM_MODEL: openai/gpt-oss-20b
366-
HINDSIGHT_API_LLM_GROQ_SERVICE_TIER: flex
364+
HINDSIGHT_API_LLM_MODEL: gpt-4o-mini
367365
GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
368366
# Prefer CPU-only PyTorch in CI (but keep PyPI for everything else)
369367
UV_INDEX: pytorch=https://download.pytorch.org/whl/cpu
@@ -417,10 +415,9 @@ jobs:
417415
test-python-client:
418416
runs-on: ubuntu-latest
419417
env:
420-
HINDSIGHT_API_LLM_PROVIDER: groq
421-
HINDSIGHT_API_LLM_API_KEY: ${{ secrets.GROQ_API_KEY }}
422-
HINDSIGHT_API_LLM_MODEL: openai/gpt-oss-20b
423-
HINDSIGHT_API_LLM_GROQ_SERVICE_TIER: flex
418+
HINDSIGHT_API_LLM_PROVIDER: openai
419+
HINDSIGHT_API_LLM_API_KEY: ${{ secrets.OPENAI_API_KEY }}
420+
HINDSIGHT_API_LLM_MODEL: gpt-4o-mini
424421
HINDSIGHT_API_URL: http://localhost:8888
425422
GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
426423
# Prefer CPU-only PyTorch in CI (but keep PyPI for everything else)
@@ -462,7 +459,6 @@ jobs:
462459
HINDSIGHT_API_LLM_PROVIDER=${{ env.HINDSIGHT_API_LLM_PROVIDER }}
463460
HINDSIGHT_API_LLM_API_KEY=${{ env.HINDSIGHT_API_LLM_API_KEY }}
464461
HINDSIGHT_API_LLM_MODEL=${{ env.HINDSIGHT_API_LLM_MODEL }}
465-
HINDSIGHT_API_LLM_GROQ_SERVICE_TIER=${{ env.HINDSIGHT_API_LLM_GROQ_SERVICE_TIER }}
466462
EOF
467463
468464
- name: Start API server
@@ -495,10 +491,9 @@ jobs:
495491
test-typescript-client:
496492
runs-on: ubuntu-latest
497493
env:
498-
HINDSIGHT_API_LLM_PROVIDER: groq
499-
HINDSIGHT_API_LLM_API_KEY: ${{ secrets.GROQ_API_KEY }}
500-
HINDSIGHT_API_LLM_MODEL: openai/gpt-oss-20b
501-
HINDSIGHT_API_LLM_GROQ_SERVICE_TIER: flex
494+
HINDSIGHT_API_LLM_PROVIDER: openai
495+
HINDSIGHT_API_LLM_API_KEY: ${{ secrets.OPENAI_API_KEY }}
496+
HINDSIGHT_API_LLM_MODEL: gpt-4o-mini
502497
HINDSIGHT_API_URL: http://localhost:8888
503498
GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
504499
# Prefer CPU-only PyTorch in CI (but keep PyPI for everything else)
@@ -545,7 +540,6 @@ jobs:
545540
HINDSIGHT_API_LLM_PROVIDER=${{ env.HINDSIGHT_API_LLM_PROVIDER }}
546541
HINDSIGHT_API_LLM_API_KEY=${{ env.HINDSIGHT_API_LLM_API_KEY }}
547542
HINDSIGHT_API_LLM_MODEL=${{ env.HINDSIGHT_API_LLM_MODEL }}
548-
HINDSIGHT_API_LLM_GROQ_SERVICE_TIER=${{ env.HINDSIGHT_API_LLM_GROQ_SERVICE_TIER }}
549543
EOF
550544
551545
- name: Start API server
@@ -578,10 +572,9 @@ jobs:
578572
test-rust-client:
579573
runs-on: ubuntu-latest
580574
env:
581-
HINDSIGHT_API_LLM_PROVIDER: groq
582-
HINDSIGHT_API_LLM_API_KEY: ${{ secrets.GROQ_API_KEY }}
583-
HINDSIGHT_API_LLM_MODEL: openai/gpt-oss-20b
584-
HINDSIGHT_API_LLM_GROQ_SERVICE_TIER: flex
575+
HINDSIGHT_API_LLM_PROVIDER: openai
576+
HINDSIGHT_API_LLM_API_KEY: ${{ secrets.OPENAI_API_KEY }}
577+
HINDSIGHT_API_LLM_MODEL: gpt-4o-mini
585578
HINDSIGHT_API_URL: http://localhost:8888
586579
GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
587580
# Prefer CPU-only PyTorch in CI (but keep PyPI for everything else)
@@ -627,7 +620,6 @@ jobs:
627620
HINDSIGHT_API_LLM_PROVIDER=${{ env.HINDSIGHT_API_LLM_PROVIDER }}
628621
HINDSIGHT_API_LLM_API_KEY=${{ env.HINDSIGHT_API_LLM_API_KEY }}
629622
HINDSIGHT_API_LLM_MODEL=${{ env.HINDSIGHT_API_LLM_MODEL }}
630-
HINDSIGHT_API_LLM_GROQ_SERVICE_TIER=${{ env.HINDSIGHT_API_LLM_GROQ_SERVICE_TIER }}
631623
EOF
632624
633625
- name: Start API server
@@ -660,10 +652,9 @@ jobs:
660652
test-go-client:
661653
runs-on: ubuntu-latest
662654
env:
663-
HINDSIGHT_API_LLM_PROVIDER: groq
664-
HINDSIGHT_API_LLM_API_KEY: ${{ secrets.GROQ_API_KEY }}
665-
HINDSIGHT_API_LLM_MODEL: openai/gpt-oss-20b
666-
HINDSIGHT_API_LLM_GROQ_SERVICE_TIER: flex
655+
HINDSIGHT_API_LLM_PROVIDER: openai
656+
HINDSIGHT_API_LLM_API_KEY: ${{ secrets.OPENAI_API_KEY }}
657+
HINDSIGHT_API_LLM_MODEL: gpt-4o-mini
667658
HINDSIGHT_API_URL: http://localhost:8888
668659
GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
669660
# Prefer CPU-only PyTorch in CI (but keep PyPI for everything else)
@@ -703,7 +694,6 @@ jobs:
703694
HINDSIGHT_API_LLM_PROVIDER=${{ env.HINDSIGHT_API_LLM_PROVIDER }}
704695
HINDSIGHT_API_LLM_API_KEY=${{ env.HINDSIGHT_API_LLM_API_KEY }}
705696
HINDSIGHT_API_LLM_MODEL=${{ env.HINDSIGHT_API_LLM_MODEL }}
706-
HINDSIGHT_API_LLM_GROQ_SERVICE_TIER=${{ env.HINDSIGHT_API_LLM_GROQ_SERVICE_TIER }}
707697
EOF
708698
709699
- name: Start API server
@@ -740,10 +730,9 @@ jobs:
740730
test-openclaw-integration:
741731
runs-on: ubuntu-latest
742732
env:
743-
HINDSIGHT_API_LLM_PROVIDER: groq
744-
HINDSIGHT_API_LLM_API_KEY: ${{ secrets.GROQ_API_KEY }}
745-
HINDSIGHT_API_LLM_MODEL: openai/gpt-oss-20b
746-
HINDSIGHT_API_LLM_GROQ_SERVICE_TIER: flex
733+
HINDSIGHT_API_LLM_PROVIDER: openai
734+
HINDSIGHT_API_LLM_API_KEY: ${{ secrets.OPENAI_API_KEY }}
735+
HINDSIGHT_API_LLM_MODEL: gpt-4o-mini
747736
HINDSIGHT_API_URL: http://localhost:8888
748737
HINDSIGHT_EMBED_PACKAGE_PATH: ${{ github.workspace }}/hindsight-embed
749738
GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
@@ -810,7 +799,6 @@ jobs:
810799
HINDSIGHT_API_LLM_PROVIDER=${{ env.HINDSIGHT_API_LLM_PROVIDER }}
811800
HINDSIGHT_API_LLM_API_KEY=${{ env.HINDSIGHT_API_LLM_API_KEY }}
812801
HINDSIGHT_API_LLM_MODEL=${{ env.HINDSIGHT_API_LLM_MODEL }}
813-
HINDSIGHT_API_LLM_GROQ_SERVICE_TIER=${{ env.HINDSIGHT_API_LLM_GROQ_SERVICE_TIER }}
814802
EOF
815803
816804
- name: Start API server
@@ -843,10 +831,9 @@ jobs:
843831
test-integration:
844832
runs-on: ubuntu-latest
845833
env:
846-
HINDSIGHT_API_LLM_PROVIDER: groq
847-
HINDSIGHT_API_LLM_API_KEY: ${{ secrets.GROQ_API_KEY }}
848-
HINDSIGHT_API_LLM_MODEL: openai/gpt-oss-20b
849-
HINDSIGHT_API_LLM_GROQ_SERVICE_TIER: flex
834+
HINDSIGHT_API_LLM_PROVIDER: openai
835+
HINDSIGHT_API_LLM_API_KEY: ${{ secrets.OPENAI_API_KEY }}
836+
HINDSIGHT_API_LLM_MODEL: gpt-4o-mini
850837
HINDSIGHT_API_URL: http://localhost:8888
851838
GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
852839
UV_INDEX: pytorch=https://download.pytorch.org/whl/cpu
@@ -903,7 +890,6 @@ jobs:
903890
HINDSIGHT_API_LLM_PROVIDER=${{ env.HINDSIGHT_API_LLM_PROVIDER }}
904891
HINDSIGHT_API_LLM_API_KEY=${{ env.HINDSIGHT_API_LLM_API_KEY }}
905892
HINDSIGHT_API_LLM_MODEL=${{ env.HINDSIGHT_API_LLM_MODEL }}
906-
HINDSIGHT_API_LLM_GROQ_SERVICE_TIER=${{ env.HINDSIGHT_API_LLM_GROQ_SERVICE_TIER }}
907893
EOF
908894
909895
- name: Start API server
@@ -994,10 +980,9 @@ jobs:
994980
test-embed:
995981
runs-on: ubuntu-latest
996982
env:
997-
HINDSIGHT_API_LLM_PROVIDER: groq
998-
HINDSIGHT_API_LLM_API_KEY: ${{ secrets.GROQ_API_KEY }}
999-
HINDSIGHT_API_LLM_MODEL: openai/gpt-oss-20b
1000-
HINDSIGHT_API_LLM_GROQ_SERVICE_TIER: flex
983+
HINDSIGHT_API_LLM_PROVIDER: openai
984+
HINDSIGHT_API_LLM_API_KEY: ${{ secrets.OPENAI_API_KEY }}
985+
HINDSIGHT_API_LLM_MODEL: gpt-4o-mini
1001986
# Prefer CPU-only PyTorch in CI
1002987
UV_INDEX: pytorch=https://download.pytorch.org/whl/cpu
1003988

@@ -1039,14 +1024,13 @@ jobs:
10391024
test-hindsight-all:
10401025
runs-on: ubuntu-latest
10411026
env:
1042-
HINDSIGHT_API_LLM_PROVIDER: groq
1043-
HINDSIGHT_API_LLM_API_KEY: ${{ secrets.GROQ_API_KEY }}
1044-
HINDSIGHT_API_LLM_MODEL: openai/gpt-oss-20b
1045-
HINDSIGHT_API_LLM_GROQ_SERVICE_TIER: flex
1027+
HINDSIGHT_API_LLM_PROVIDER: openai
1028+
HINDSIGHT_API_LLM_API_KEY: ${{ secrets.OPENAI_API_KEY }}
1029+
HINDSIGHT_API_LLM_MODEL: gpt-4o-mini
10461030
# For test_server_integration.py compatibility
1047-
HINDSIGHT_LLM_PROVIDER: groq
1048-
HINDSIGHT_LLM_API_KEY: ${{ secrets.GROQ_API_KEY }}
1049-
HINDSIGHT_LLM_MODEL: openai/gpt-oss-20b
1031+
HINDSIGHT_LLM_PROVIDER: openai
1032+
HINDSIGHT_LLM_API_KEY: ${{ secrets.OPENAI_API_KEY }}
1033+
HINDSIGHT_LLM_MODEL: gpt-4o-mini
10501034
# Prefer CPU-only PyTorch in CI
10511035
UV_INDEX: pytorch=https://download.pytorch.org/whl/cpu
10521036

@@ -1089,10 +1073,9 @@ jobs:
10891073
runs-on: ubuntu-latest
10901074
needs: test-rust-cli
10911075
env:
1092-
HINDSIGHT_API_LLM_PROVIDER: groq
1093-
HINDSIGHT_API_LLM_API_KEY: ${{ secrets.GROQ_API_KEY }}
1094-
HINDSIGHT_API_LLM_MODEL: openai/gpt-oss-20b
1095-
HINDSIGHT_API_LLM_GROQ_SERVICE_TIER: flex
1076+
HINDSIGHT_API_LLM_PROVIDER: openai
1077+
HINDSIGHT_API_LLM_API_KEY: ${{ secrets.OPENAI_API_KEY }}
1078+
HINDSIGHT_API_LLM_MODEL: gpt-4o-mini
10961079
HINDSIGHT_API_URL: http://localhost:8888
10971080
GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
10981081
UV_INDEX: pytorch=https://download.pytorch.org/whl/cpu
@@ -1148,7 +1131,6 @@ jobs:
11481131
HINDSIGHT_API_LLM_PROVIDER=${{ env.HINDSIGHT_API_LLM_PROVIDER }}
11491132
HINDSIGHT_API_LLM_API_KEY=${{ env.HINDSIGHT_API_LLM_API_KEY }}
11501133
HINDSIGHT_API_LLM_MODEL=${{ env.HINDSIGHT_API_LLM_MODEL }}
1151-
HINDSIGHT_API_LLM_GROQ_SERVICE_TIER=${{ env.HINDSIGHT_API_LLM_GROQ_SERVICE_TIER }}
11521134
EOF
11531135
11541136
- name: Start API server
@@ -1183,10 +1165,9 @@ jobs:
11831165
test-upgrade:
11841166
runs-on: ubuntu-latest
11851167
env:
1186-
HINDSIGHT_API_LLM_PROVIDER: groq
1187-
HINDSIGHT_API_LLM_API_KEY: ${{ secrets.GROQ_API_KEY }}
1188-
HINDSIGHT_API_LLM_MODEL: openai/gpt-oss-20b
1189-
HINDSIGHT_API_LLM_GROQ_SERVICE_TIER: flex
1168+
HINDSIGHT_API_LLM_PROVIDER: openai
1169+
HINDSIGHT_API_LLM_API_KEY: ${{ secrets.OPENAI_API_KEY }}
1170+
HINDSIGHT_API_LLM_MODEL: gpt-4o-mini
11901171
GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
11911172
UV_INDEX: pytorch=https://download.pytorch.org/whl/cpu
11921173

‎CLAUDE.md‎

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -317,7 +317,7 @@ npm install
317317
Required env vars:
318318
- `HINDSIGHT_API_LLM_PROVIDER`: openai, anthropic, gemini, groq, ollama, lmstudio
319319
- `HINDSIGHT_API_LLM_API_KEY`: Your API key
320-
- `HINDSIGHT_API_LLM_MODEL`: Model name (e.g., o3-mini, claude-sonnet-4-20250514)
320+
- `HINDSIGHT_API_LLM_MODEL`: Model name (e.g., gpt-4o-mini, claude-sonnet-4-20250514)
321321

322322
Optional (uses local models by default):
323323
- `HINDSIGHT_API_EMBEDDINGS_PROVIDER`: local (default) or tei

‎docker/test-image.sh‎

Lines changed: 10 additions & 10 deletions
Original file line numberDiff line numberDiff line change
@@ -13,9 +13,9 @@
1313
# target - Optional: 'cp-only' for control plane, otherwise assumes API image (default: api)
1414
#
1515
# Environment variables:
16-
# GROQ_API_KEY - Required for API/standalone images (LLM verification)
17-
# HINDSIGHT_API_LLM_PROVIDER - LLM provider (default: groq)
18-
# HINDSIGHT_API_LLM_MODEL - LLM model (default: llama-3.3-70b-versatile)
16+
# HINDSIGHT_API_LLM_API_KEY - Required for API/standalone images (LLM verification)
17+
# HINDSIGHT_API_LLM_PROVIDER - LLM provider (default: openai)
18+
# HINDSIGHT_API_LLM_MODEL - LLM model (default: gpt-4o-mini)
1919
# HINDSIGHT_API_EMBEDDINGS_PROVIDER - Embeddings provider (optional, for slim images: openai, cohere, tei)
2020
# HINDSIGHT_API_EMBEDDINGS_OPENAI_API_KEY - OpenAI API key for embeddings (optional)
2121
# HINDSIGHT_API_RERANKER_PROVIDER - Reranker provider (optional, for slim images: cohere, tei)
@@ -34,7 +34,7 @@
3434
# ./docker/test-image.sh hindsight-control-plane:test cp-only
3535
#
3636
# # Test slim image with external providers
37-
# export GROQ_API_KEY=gsk_xxx
37+
# export HINDSIGHT_API_LLM_API_KEY=sk_xxx
3838
# export HINDSIGHT_API_EMBEDDINGS_PROVIDER=openai
3939
# export HINDSIGHT_API_EMBEDDINGS_OPENAI_API_KEY=sk-xxx
4040
# export HINDSIGHT_API_RERANKER_PROVIDER=cohere
@@ -60,8 +60,8 @@ IMAGE="${1:-}"
6060
TARGET="${2:-api}"
6161
TIMEOUT="${SMOKE_TEST_TIMEOUT:-120}"
6262
CONTAINER_NAME="${SMOKE_TEST_CONTAINER_NAME:-hindsight-smoke-test}"
63-
LLM_PROVIDER="${HINDSIGHT_API_LLM_PROVIDER:-groq}"
64-
LLM_MODEL="${HINDSIGHT_API_LLM_MODEL:-llama-3.3-70b-versatile}"
63+
LLM_PROVIDER="${HINDSIGHT_API_LLM_PROVIDER:-openai}"
64+
LLM_MODEL="${HINDSIGHT_API_LLM_MODEL:-gpt-4o-mini}"
6565

6666
# Validate arguments
6767
if [ -z "$IMAGE" ]; then
@@ -88,9 +88,9 @@ else
8888
fi
8989

9090
# Check for required environment variables
91-
if [ "$NEEDS_LLM" = true ] && [ -z "${GROQ_API_KEY:-}" ]; then
92-
echo -e "${RED}Error: GROQ_API_KEY environment variable is required for API/standalone images${NC}"
93-
echo "Set it with: export GROQ_API_KEY=your-api-key"
91+
if [ "$NEEDS_LLM" = true ] && [ -z "${HINDSIGHT_API_LLM_API_KEY:-}" ]; then
92+
echo -e "${RED}Error: HINDSIGHT_API_LLM_API_KEY environment variable is required for API/standalone images${NC}"
93+
echo "Set it with: export HINDSIGHT_API_LLM_API_KEY=your-api-key"
9494
exit 2
9595
fi
9696

@@ -123,7 +123,7 @@ else
123123
# Build docker run command with required and optional env vars
124124
DOCKER_CMD="docker run -d --name $CONTAINER_NAME"
125125
DOCKER_CMD="$DOCKER_CMD -e HINDSIGHT_API_LLM_PROVIDER=$LLM_PROVIDER"
126-
DOCKER_CMD="$DOCKER_CMD -e HINDSIGHT_API_LLM_API_KEY=${GROQ_API_KEY}"
126+
DOCKER_CMD="$DOCKER_CMD -e HINDSIGHT_API_LLM_API_KEY=${HINDSIGHT_API_LLM_API_KEY}"
127127
DOCKER_CMD="$DOCKER_CMD -e HINDSIGHT_API_LLM_MODEL=$LLM_MODEL"
128128

129129
# Add optional embeddings provider config

‎docker/test-slim-local.sh‎

Lines changed: 5 additions & 9 deletions
Original file line numberDiff line numberDiff line change
@@ -6,24 +6,17 @@
66
# It expects API keys to be set in environment variables.
77
#
88
# Usage:
9-
# export GROQ_API_KEY=gsk_xxx
109
# export OPENAI_API_KEY=sk-xxx
1110
# export COHERE_API_KEY=xxx
1211
# ./docker/test-slim-local.sh
1312
#
1413
# Or inline:
15-
# GROQ_API_KEY=gsk_xxx OPENAI_API_KEY=sk_xxx COHERE_API_KEY=xxx ./docker/test-slim-local.sh
14+
# OPENAI_API_KEY=sk_xxx COHERE_API_KEY=xxx ./docker/test-slim-local.sh
1615
#
1716

1817
set -euo pipefail
1918

2019
# Check for required API keys
21-
if [ -z "${GROQ_API_KEY:-}" ]; then
22-
echo "❌ Error: GROQ_API_KEY environment variable is required"
23-
echo "Set it with: export GROQ_API_KEY=gsk_xxx"
24-
exit 1
25-
fi
26-
2720
if [ -z "${OPENAI_API_KEY:-}" ]; then
2821
echo "❌ Error: OPENAI_API_KEY environment variable is required"
2922
echo "Set it with: export OPENAI_API_KEY=sk-xxx"
@@ -41,7 +34,10 @@ IMAGE="${1:-hindsight-slim:test}"
4134
echo "Testing image: $IMAGE"
4235
echo ""
4336

44-
# Set up external providers
37+
# Set up LLM and external providers
38+
export HINDSIGHT_API_LLM_PROVIDER=openai
39+
export HINDSIGHT_API_LLM_API_KEY=$OPENAI_API_KEY
40+
export HINDSIGHT_API_LLM_MODEL=gpt-4o-mini
4541
export HINDSIGHT_API_EMBEDDINGS_PROVIDER=openai
4642
export HINDSIGHT_API_EMBEDDINGS_OPENAI_API_KEY=$OPENAI_API_KEY
4743
export HINDSIGHT_API_RERANKER_PROVIDER=cohere

‎hindsight-api/hindsight_api/config.py‎

Lines changed: 2 additions & 2 deletions
Original file line numberDiff line numberDiff line change
@@ -314,7 +314,7 @@ def normalize_config_dict(config: dict[str, Any]) -> dict[str, Any]:
314314

315315
# Provider-specific default models
316316
PROVIDER_DEFAULT_MODELS = {
317-
"openai": "o3-mini",
317+
"openai": "gpt-4o-mini",
318318
"anthropic": "claude-haiku-4-5-20251001",
319319
"gemini": "gemini-2.5-flash",
320320
"groq": "openai/gpt-oss-120b",
@@ -325,7 +325,7 @@ def normalize_config_dict(config: dict[str, Any]) -> dict[str, Any]:
325325
"claude-code": "claude-sonnet-4-5-20250929",
326326
"mock": "mock-model",
327327
}
328-
DEFAULT_LLM_MODEL = "o3-mini" # Fallback if provider not in table
328+
DEFAULT_LLM_MODEL = "gpt-4o-mini" # Fallback if provider not in table
329329
DEFAULT_LLM_MAX_CONCURRENT = 32
330330
DEFAULT_LLM_MAX_RETRIES = 10 # Max retry attempts for LLM API calls
331331
DEFAULT_LLM_INITIAL_BACKOFF = 1.0 # Initial backoff in seconds for retry exponential backoff

‎hindsight-api/tests/test_provider_default_models.py‎

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -93,7 +93,7 @@ def test_per_operation_provider_default_model():
9393
config = HindsightConfig.from_env()
9494

9595
# Global LLM should use OpenAI default
96-
assert config.llm_model == "o3-mini", f"Expected o3-mini, got {config.llm_model}"
96+
assert config.llm_model == "gpt-4o-mini", f"Expected gpt-4o-mini, got {config.llm_model}"
9797

9898
# Retain should use Anthropic default
9999
assert (

0 commit comments

Comments
 (0)